Article

Urban Studies 2015, Vol. 52(1) 20–49 Ó Urban Studies Journal Limited 2014 Industrial gentrification in West Reprints and permissions: sagepub.co.uk/journalsPermissions.nav Chelsea, New York: Who survived DOI: 10.1177/0042098014536785 and who did not? Empirical evidence usj.sagepub.com from discrete-time survival analysis

Heeyeun Yoon Harvard University Graduate School of Design, USA

Elizabeth Currid-Halkett University of Southern California, USA

Abstract This paper empirically tests the extent to which economic restructuring and gentrification affect viability and vulnerability businesses, with specific focus on arts and cultural industries in West Chelsea from 2000 through 2012. Based on the theoretical framework, gentrification stage model and adopting discrete-time survival analysis, we separately compare the risks of opening and clos- ing between businesses established before/early stage of revitalisation (early-arrivers) and those established in the later stage (late-arrivers) within West Chelsea, versus their counterparts in the remainder of the study area in New York. We find that West Chelsea has been an advantageous location overall for late-arrivers in surviving in their market, while the early-arrived gallery and individual artists’ enterprises have faced a higher risk of their operations closing. On the other hand, a higher proportion of new gallery and arts and cultural industries remain attracted to West Chelsea after 2000, suggesting that firms in those industries may be benefiting from the agglomeration effects and localisation economies associated with colocation. The higher opening probability of lodging venues (e.g. hotels) and other amenities signals an overall transformation of the neighbourhood and influx of new uses (and visitors) observed during this time frame.

Keywords arts and cultural industry, gentrification, industrial restructuring, survival analysis, West Chelsea

Received December 2013; accepted April 2014

Introduction Economic restructuring or industrial gentri- fication is an ongoing discussion in the geo- Corresponding author: graphical and social science literature. Elizabeth Currid-Halkett, University of Southern California, School of Public Policy, 650 Child’s Way, Los Angeles, CA Throughout history, new industries have 90089, USA. emerged in areas formerly occupied by less Email: [email protected] Yoon and Currid-Halkett 21 competitive ones resulting in a transforma- stratum of the new middle class) (Ley, tion of the social, cultural and economic geo- 1996). graphy (Curran, 2007; Giloth and Betancur, However, there has not been a consensus 1988; Zukin, 1989). Particularly in the post- that these industries are demonstrative of industrial era, city officials seek replacements the final stage of economic restructuring, for once-dominant, but lost manufacturing and whether their role is one of victim or vic- industries to maintain fiscal viability tor in the gentrification process. Stage theory (Fainstein, 1999; Fainstein and Judd, 1999; of demographic gentrification hypothesises Zukin, 1989). Managerial operations, legal that artists’ groups and cultural producers services, and finance, insurance and real are often pioneers in gentrification, and estate consulting are some of the industries actively participate in reviving urban econo- that economic development practitioners mies, but, as Zukin’s seminal work on ‘loft tried to recruit in the wake of dramatic glo- living’ documents, their geographic territory bal economic restructuring (Drennan, 1991; is not safe from subsequent phase of gentrifi- Hudson, 1984; Hyra, 2008; Sassen, 1994). cation in a longer term (Watt, 2005; Zukin, Demographic changes also accompany the 1989). A similar theory applies to industrial appearance of these service industries, intro- gentrification process broadly speaking. The ducing middle class and white-collar workers revitalisation of SoHo, Greenwich Village who could provide an improved tax base to and the East Village in New York City are the city (Beauregard, 2009; Clay, 1979; Rose, known widely as such examples, demonstrat- 1984; Smith and Defilippis, 1999). ing the vulnerability of the arts and cultural Creative, artistic and cultural industries industries amid active neighbourhood revita- are increasingly being recognised as a sup- lisation and urban economic change in the plement or an equivalent urban economic postindustrial era (Abu-Lughod, 1994; base to the aforementioned industries, both Beard and Berlowitz, 1993; Hudson, 1984, as a part of the ‘creative class’ leitmotif and 1987; Mele, 2000; Simpson, 1981; Smith and as a central force in and of themselves. Such DeFilippis,1999; Soffer, 2010; Zukin, 1989; industries have indeed contributed a robust Zukin and Braslow, 2011). share of values that cities generate (Currid, West Chelsea has been deemed a succes- 2006; Markusen and Schrock, 2006; sor to these neighbourhoods because of its Markusen et al. 2004), and have also created recent agglomeration of art galleries, artists’ unique environmental settings that further live-work spaces, and the cultural atmo- lure residents and visitors (Florida, 2002, sphere, after its long history as a manufac- 2005; Hutton, 2009). Unlike white-collar turing core in New York City (Fainstein, middle classes associated with managerial, 2007; Molotch and Treskon, 2009). capitalist services or ‘the organisation man’ Celebrating and further enhancing this trans- (Whyte, 1956), members of the arts and formation, New York City’s Department of cultural industries are characterised by City Planning (DCP) rezoned West Chelsea their unique cultural inclinations, educa- to protect the gallery district while promot- tion pedigree and entrepreneurial ethos ing general revitalisation of the area (West (Florida 2002; Lloyd, 2006, among others). Chelsea rezoning). While this public inter- This new urban elite has been anointed vention has been highly regarded for its many titles: (a subset of ) the Super- innovativeness (DCP, 2006b), some recent Creative Core (Florida, 2005), the bobos anecdotal evidence of commodification sug- (Brooks, 2010), the new cultural class (as a gests that artistic enterprises, and their 22 Urban Studies 52(1) companion industries, are now being priced Using a unique methodological out (Brozan, 2004; Kazakina, 2013; Moss, approach and new data set, we empirically 2012a; Ouroussoff, 2006; Wilson, 2005). test the extent to which economic restruc- A natural assumption is that West turing and gentrification affect viability Chelsea is undergoing the repeated pattern and vulnerability of the industries, with shown in SoHo, Greenwich Village and the specific focus on the arts and cultural East Village neighbourhoods, becoming industries in West Chelsea from 2000 another example of arts and cultural indus- through 2012, as they fared from the outset tries as a significant variable in the beginning of the process. With discrete-time survival phase of neighbourhood economic restruc- analysis, we measure the risk of closing turing. Yet, a competing hypothesis may be business and probability of businesses that West Chelsea has firmly stabilised itself opening pertaining to arts and cultural as a creative, cultural cluster in New York industries that entered West Chelsea at two City and will remain as so despite ongoing stages in the development process. economic restructuring and gentrification of Comparisons will be made in between the neighbourhood. Put another way, is it groups of such establishments that arrived possible for a neighbourhood to remain an to the neighbourhood in early and later artistic and cultural centre even after gentrifi- phases of gentrification, versus their coun- cation and redevelopment set in? terparts’ arrival in a reference area defined Extensive studies have explored the trans- in Manhattan during those two different formation of postindustrial neighbourhoods timeframes. Other non-arts and cultural through the lens of artists and the cultural industries will be also analysed in the same industries (Abu-Lughod, 1994; Beard and format, to aid general understanding of the Berlowitz, 1993; Zukin, 1989, among others). neighbourhood’s postindustrial economic Recent interest in West Chelsea’s transforma- landscape. tion from industrial to artistic and postindus- trial neighbourhood has not yet prompted sufficient empirical studies. West Chelsea Background and context deserves particular attention since it repre- Theoretical background: Stage models of sents one of the latest examples of such phe- nomena and one of the rare examples where gentrification active policy intervention aimed to preserve Residential gentrification is often explained the arts and cultural industries from displace- as stages or successions of a neighbourhood ment. How have these industries fared, as a revitalisation cycle, from inception to final neighbourhood has been significantly equilibrium (Kerstein, 1990; Lees, 2000; Ley impacted by economic restructuring and gen- 2003; Smith, 2012). Despite considerable trification? Like SoHo, the arts and cultural diversity in types and phasing under this industries were actively cultivated as early concept, the basic notion identifies differ- regenerators, but does the story of West ences in in-movers’ socioeconomic and cul- Chelsea mimic that of its predecessor or does tural dimensions based on their arrival it present an alternative trajectory for pioneer- timing to the subject neighbourhood. ing artistic neighbourhoods? Also, how has Generally, earlier arrivers are believed to be the policy influenced the business landscape ‘risk oblivious’, as they gravitate to an obso- and economic geography of West Chelsea lete neighbourhood and transform it to during the course? Could we learn something inhabitable shape. Later arrivers are ‘risk new from the case of West Chelsea? averse’, who tend to migrate to the Yoon and Currid-Halkett 23 neighbourhood when it is substantially businesses from the art-related to commer- improved (Gale, 1979; Kerstein, 1990). cial enterprises such as cafe´ and restaurants Later arrivers are more likely to be the and media play a role to increase the visibi- traditional middle classes represented by lity of such changing neighbourhoods. managerial and capitalist service that Consequently the original arty enclaves requires advanced education attainment and become neither attractive nor affordable to offers higher incomes. Earlier arrivers, in the frontiers anymore, while whichever busi- contrast, have relatively modest economic nesses that could pay higher rent displace means and are much more diverse in their them eventually (Lloyd, 2006; Zukin, 2010; social and occupational characteristics, so- Zukin and Braslow, 2011). called ‘richer in cultural capital than eco- Wealthier residents also appear as gentri- nomic capital’, borrowing Bourdieu’s defini- fiers in the later stage of industrial gentrifi- tion (Kerstein, 1990; Smith, 2002; Van cation. Real estate developers have paid Criekingen and Decroly, 2003). Rose’s attention to this type of emerging art cluster (1996, 1984) case studies of Canadian cities to make a profit from the currently underva- and Bondi’s (1999) Edinburgh research iden- lued but soon-to-be appreciated land, and tified ‘marginal professionals’ as early- local governments are willing to allow them arrivers whose choice of location was largely to increase housing stocks and ignite neigh- lifestyle oriented. The authors observed that bourhood revitalisation (Zukin, 1989; Zukin they worked for fragmented professions and Braslow, 2011). In the heated real estate related to art or public affairs. market, property owners get a windfall gain The invasion of the frontiers is critical in while renters – either residents or businesses the early stage of the gentrification process, – suffer from the threat of displacement. but their duration is challenged (Watt, 2005; Recent discourses began to view creative, Zukin, 1989). Smith (2012), Smith and cultural industries and their individuals as a DeFelippis (1999) and Cameron and Coaffee more robust and established economic force (2005) argued that these people come to for- who sustain themselves throughout the mer sweat equities to ‘aestheticise’ the neigh- whole phase of gentrification and substan- bourhood (Ley, 2003) then give way to more tially contribute to city’s economy. Currid affluent gentrifiers when these places become (2006), Markusen et al. (2004) and excessively valorised. Markusen and Schrock (2006) found that a A similar cycle is found in industrial gen- greater concentration of creative and cul- trification. The frontier industries, notably tural employments explain better economic art and cultural industries, move to urban health of the cities. The Center for an Urban areas that used to be occupied by manufac- Future (2011) described the disproportionate turing industries (Philip Habib & Associates growth of design and art enterprises in New (PHA), 2005). Frontiers’ quintessential cul- York City that filled the fiscal void left by ture appeals to gentrifier industries. For manufacturing sectors. tourism enterprises, the image of this cul- These industries are also acclaimed for ture, so-called bohemian milieu, is a good their lead in setting up a business environ- ingredient to commodify the place to attract ment to further attract other industries visitors who seek unique and authentic (Florida, 2002, 2005; Hutton, 2009; Lloyd, experiences (Fainstein and Judd, 1999), and 2006; Markusen and Schrock, 2006). want to ‘consume’ the arts (Zukin and Florida’s (2002) research on 50 US Braslow, 2011: 132). Realising this phenom- Metropolitan Statistical Areas revealed that enon, some entrepreneur artists expand their higher ‘bohemian index’– quantified cultural 24 Urban Studies 52(1) milieu for artist cohorts, and the number of the 1960s, by accommodating in-migration high-tech industries and their talented of artists in its abundant derelict mills. West human capital – were likely to coexist, Chelsea underwent a similar cycle from the although causal relationship could not be late 1980s, by initially hosting art collectors corroborated. Lloyd (2006), in his chronol- and dealers who moved from SoHo, then ogy of industrial restructuring in Chicago’s further attracting artists groups (Holusha, Wicker Park, observed relevance of artist 1997; Morgan, 2012; PHA, 2005), ultimately workforces’ presence in 1990s in explaining becoming an art gallery district1 (NYLMPC, the current agglomeration of high-tech 2008). enterprises. West Chelsea in the postindustrial era. Two Historic background institutional efforts weighed in Chelsea’s postindustrial evolvement; Chelsea rezoning West Chelsea in the industrial era. West (1999) and West Chelsea rezoning (2005), Chelsea, a former manufacturing core, is a along with the larger efforts to revive the newly emerged cluster of arts and cultural west side of Midtown. The Chelsea Plan pro- industry in New York City that has gained posed by the Community Board 4 in 1996 publicity since the 1990s. It is on the west and Far West Midtown – a framework for side of 10th avenue in Chelsea, known as a development conceived by the Department neighbourhood bounded by W14th Street to of City Planning in 2001 included projects the south, W34th Street to the north, and such as Hudson Yard development, the 6th Avenue to the east, and bordered by extension of Number 7 Line and Hudson Hudson River waterfront to the west (DCP, Park & Boulevard (DCP, 1996, 2001). While 1999; NYbits, n.d.). Chelsea rezoning mainly concerned the East While New York City restructured itself side of Chelsea in accommodating demand to favour producer services in response to a for non-manufacturing spaces, it also opened global deindustrialisation process in Western up an official consideration on West Chelsea cities, the progress of the postindustrial by up-zoning one and a half blocks of for- transformation of West Chelsea could not mer manufacturing land (DCP, 2006a). keep pace with other neighbourhoods, The West Chelsea rezoning of 2005 mainly because of its long-standing designa- focused specifically on 13 blocks and two tion as a manufacturing district (PHA, half-blocks of West Chelsea (Special West 2005). Without embracing an influx of pro- Chelsea district, roughly bounded by W16th fessional service industries or new residents, Street to the south, W30th Street to the it has remained as a shell of light and uncon- north, 10th and 11th Avenues to the east solidated manufacturing including meat- and west, respectively). The primary purpose packing and auto repair until the late 1980s of this rezoning was twofold: first, to (David and Hammond, 2011; New York encourage mixed-use development of the City Landmarks Preservation Commission neighbourhood; second, to advance the (NYLMPC), 2008). development and use of the High Line as a West Chelsea’s revival followed SoHo’s public park. wake (Molotch and Treskon, 2009; To institute mixed-used development, NYLMPC, 2008). Like West Chelsea, SoHo land use change and density increment con- also had been a historic manufacturing and stituted a large part of the rezoning, in that commercial district before it began to gain more than six blocks were rezoned from the current reputation as an art district from manufacturing to commercial use with Yoon and Currid-Halkett 25 increased floor area ratio (FAR). Protecting since then, hypothetically. Unlike the former and promoting art-related business was stages, it is less clear to discern the nature of another emphasis. Non-commercial art exhi- the later arrivers, their precise timing and bition and gallery use continued to be per- type of industry. mitted as-of-right in the remaining In defining a starting point of the later manufacturing areas, where open floor plans phase, it is critical to understand those pub- and high ceilings of industrial venues were lic interventions as milestones. Even the well suited for exhibitions, but less appealing transformation of West Chelsea is perceived to any other uses. In areas concentrated with as a gradual process at the beginning, the existing galleries, such as between W20th series of rezoning and public projects appar- Street to W22nd Street, and between W24th ently took the pre-existing, largely sponta- Street and W27th Street bounded by 10th neous revitalisation process to another level. and 11th Avenues, land use was not changed The Chelsea rezoning of 1999 had created a but remained as manufacturing districts, so pro-development tone in overall Chelsea art galleries could keep taking advantage of and implemented a pilot action in West such architectural characteristics (PHA, Chelsea (DCP, 1996, 1999). The following 2005). In some portions of the newly estab- West Chelsea rezoning of 2005 institutiona- lished commercial district, non-commercial lised the redevelopment of West Chelsea, gallery use was still encouraged on the first further intensified the economic restructur- floor of buildings (DCP, 2006a). ing already in progress (Dawid, 2012; DCP, The High Line, former West Side Freight 2006a; Moss, 2012a, 2012b). Line, was converted to a park. After 30 years Given this backdrop, it is reasonable to of abandonment across West Chelsea and establish the year 2000 as a tipping point Meatpacking district, the city government toward active economic restructuring/gentri- announced to pursue the use of the High fication, subsequently to define businesses Line as a public amenity in 2004 (excluding that opened before and after 2000 as proxies the last section of the park), and completed for earlier and later arrivers, respectively. If the first and the second sections in 2009 and this assumption is valid, businesses that 2011, respectively. Between its conception opened after 2000 would differ from their and completion, the project attracted specu- forerunners with respect to industrial types, lators’ attention and induced an unprece- target customers and financial prowess. dented amount of property development in Businesses opened after 2000 may have been West Chelsea – US$2 billion of private better equipped to survive in the upwardly investment as of 2011 (McGeehan, 2011). shifting neighbourhood, because they could predict the fate of West Chelsea when they Research design began. However, businesses that opened before 2000, who did not expect such trans- Research introduction formation of the neighbourhood, were not Based on the concept of the stage model, necessarily of types and scales befitting the postindustrial restructuring of West Chelsea current West Chelsea. can be understood as a two-phase process; To understand the business landscape of first, from the 1990s when frontier arts and postindustrial West Chelsea, the gentrifica- cultural industries surfaced in a spontaneous tion process and its developmental impacts, manner (PHA, 2005), second when more we compare the risk of displacement/reloca- diverse types of later arrivers are believed to tion and probability of opening between the alter the business landscape of West Chelsea arts and cultural industries within West 26 Urban Studies 52(1)

Chelsea and their counterparts in other parts involuntarily. Or if the revitalisation leads to of Manhattan. The comparisons are made a disproportionally large amount of new separately for three sets of subsample property development or renovation in the grouped by opening years as proxies of neighbourhood, firms in old venues have to early-arrivers and late-arrivers, under the move out during the construction. Therefore assumption that their viabilities and vulner- in this neighbourhood, the probability of abilities differ. We specifically ask: business closure may be larger than that of average firms in the same industry in other (1) How did the active neighbourhood parts of the city. Also this threat may hit revitalisation process in West Chelsea harder or appeal to specific industries (from 2000 to 2012) affect the occur- unevenly, thus the industry composition in rence of displacement/relocation of the neighbourhood would change (Curran, later-arrived business establishments in 2007). the West Chelsea, compared with that Obviously there are other reasons for of business establishments in the cho- businesses in gentrifying West Chelsea to sen reference area of Manhattan? voluntarily end their operations, other than (2) Likewise, how does the occurrence of involuntary displacement or cessation. business displacement/relocation differ However, if there was not an effect of gentri- for early-arrivers in West Chelsea from fication, and there was only a voluntary and that of their counterparts in the refer- non-location related reasons, the probability ence area? of business closure or new opening would (3) What is the equivalent difference in the not differ in the same industry located in probabilities of opening businesses West Chelsea versus other parts of the city. (later arrivers) in the West Chelsea as In Figure 1, we describe subsamples and compared with the reference area in events of interest – closure or opening – for Manhattan? three subsequent analyses, and a brief time- line of planning interventions in Chelsea. The analytical unit is business establishment. We made use of discrete-time survival The events we investigate are closing busi- analysis as a primary analytical tool. The ness establishments in the current location, analysis investigates whether a member of a including displacement, relocation and ter- sample experiences a designated event in a mination (closure hereafter), and opening particular time point, and if so how long new business establishments. the member survived before experiencing the In analysing business gentrification in event (Singer and Willett, 1993; Yamaguchi, West Chelsea by observing the incidents of 1991). The outcome of the analysis is business closure and opening, we make two expressed as a conditional probability of sur- assumptions. First, we assume that if a vival, meaning not experiencing the event in neighbourhood undergoes revitalisation and a specific time point in the observation becomes gentrified, firms within its bound- period, only under the condition that they ary would exhibit a systematic change in have not experienced the event before. This their operation. If the existing firms could conditionality enables an unbiased estimate not afford the increased rent and operating of the population hazard probability in cer- cost, or their revenues fall because of loss of tain points of time, even in case the observa- their customer base in such an upwardly tion period is too short to fully record the shifted neighbourhood, then they may close time to events for every member of the sam- their operations from the current location ple, i.e. the presence of left-censored and Yoon and Currid-Halkett 27

Figure 1. Subsample selection for Analysis 1, 2 and 3, and policy interventions and projects implemented in West Chelsea.

right-censored data (Aalen et al., 2008; Gansevoort Street to the south, 9th Avenue Allison, 1984; Cox, 1972; Singer and Willett, to the east and the Hudson River waterfront 1991). This is possible because the probabil- to the west. We previously analysed the core ity is concerned based on the survivors from area and the West Chelsea boundary that we the event renewed in each point of the obser- currently adopt separately, and the results vation period, so-called risk set, meaning the from both are not much discernable, helping members remaining in the sample and face us to conclude that the latter depicts the the risk again in the next time point. Below extent of the location effect made by West we outline the site boundaries, data set and Chelsea rezoning, the High Line project and methodologies. their spillover effect reasonably. The analyti- cal result with the core West Chelsea will be available upon request to the authors. Site For the reference site, we have established In this study, both arts and cultural and a study area to include the south part of non-arts and cultural industries in West Manhattan, below 59th Street, at the south- Chelsea are compared with their counter- ern edge of Central Park (the study area). In parts in the reference area in Manhattan. order to distinguish the location effect of For the sake of the analysis, we define West West Chelsea, this study could have taken Chelsea (WCh) to encompass the core area, the entire Manhattan as a reference site in rezoned by West Chelsea rezoning and the order to make the control group represent a extended area to include the entire span of broad average of Manhattan. However, the the High Line to capture its spillover effects. area above and two sides of Central Park – The West Chelsea analysed in this study Harlem, Upper West and East Side, respec- is bounded by W34th Street to the north, tively – are not normally considered as 28 Urban Studies 52(1)

Figure 2. Locations of West Chelsea (WCh), the reference area, the High Line, and the boundary indicating areas of Chelsea Rezoning and West Chelsea Rezoning considerations. active commercial real estate market (Best Data set and sample Manhattan Offices, 2013; Costar Group, The primary data sets are Info USA n.d.; Reis, n.d.), and their business pattern Business List collected in 2000, and the and trend change may be too discrete from ESRI Business Location data collected the lower part of Manhattan, which would between 2006 and 2012 to form a list of make the comparison result overstated. addresses that all business establishments Therefore, the study area is limited within have resided in the study area from 2000 to the southern part of Manhattan. In Figure 2, we illustrate the site bound- 2012. These data sets are collected every aries of the study area, the West Chelsea year, recording business address, classifica- (WCh), affected area by Chelsea rezoning tion information (North American Industry and West Chelsea rezoning, and the location Classification System (NAICS)), names, of the High Line. geo-code and phone numbers. As these data Yoon and Currid-Halkett 29 sets are rigorously checked and updated Property, LLC, Cortera Plus TM, Bizapedia, every year, commercially used for the pur- Health Grades, Inc., Avvo, Inc., supplemen- pose of business location strategy, address ted with direct phone call inquiries. For information is fairly reliable. reverse look-up, we used LoopNet, Inc., From this pool of addresses, we took an DOB Search from Concert Technologies Inc. additional step for better accuracy. Rather Longitudinal sample data set is finalised than using the population data, we drew a with a total size of 17,054 in the study area, random sample of addresses and cross- within which 4184 are included in West checked the business location, opening year Chelsea boundaries. Full-year data are avail- and their current status. able for all other years, but data for 2012 The initial sample was 25,000 with over- only cover the first half of the year. The sampling in West Chelsea and some minor incomplete data for 2012 has a minor effect industries to address potential issues posed on Analysis 1 because the risk set for 2012 by simple random sampling. First, since becomes slightly smaller than what it actu- West Chelsea covers only a small fraction of ally should be, hence the probability esti- the overall study area (4.8% of study area, mates of the year 2012 are greater than the as measured in ArcGIS 10), these areas actual magnitude. While this does not affect would hold a small-sized sample as a conse- Analysis 2 at all, Analysis 3 becomes vulner- quence. And there is also a possibility that a able to bias because the opening is the event disproportionately high number of samples of interest. Therefore, the year 2012 will be may have been dropped in particular indus- excluded from the risk set of Analysis 3, lim- tries because of the absence of status record iting the observation period to 2011. resulting from the causal manner in business registration. These circumstances together would negatively affect the precision of the Measures statistical analysis and make oversampling The observation period is from 2000 to 2012 inevitable, accordingly we applied two post- for the first two analyses, and 2012 is stratification weighting in selective analyses. excluded in the third analysis. The time After this procedure, the only missing infor- dimension differs in Analysis 1 from mation is the businesses with both opening Analyses 2 and 3. In Analysis 1, we examine and closing falling within 2001 to 2005. For the probability of the event – closing busi- those 274 addresses that have incomplete list ness establishments – by age of business of business that they had accommodated establishments, while in the Analyses 2 and between 2000 and 2012, we reversely traced 3, the probability of the event – closing back the businesses name from the address. (Analysis 2) and opening (Analysis 3) – is Expanding this address list to an individ- measured by calendar year. ual business list, we then inspected and Outcome variable is log-odds of time- cleaned the data set to verify their opening invariant dichotomy, h, indicating a prob- year, current status of whether the establish- ability a business experiences the event in ments were indeed active or closed. To interest. Odds are one way to express the obtain business duration information, we probability. In this analysis, for example, referenced the Corporation & Business odds of closing business establishment ver- Entity Database of New York Department sus not doing so is explained as a ratio of of State, Division of Corporations, State the probability that a business establishment Records & UCC, the database from Manta will close operation to the probability that it Media, Inc., and Yellow Page Intellectual will not do so. 30 Urban Studies 52(1)

Question predictor is also dichotomy, from the ‘Arts, Entertainment, and WCh, indicating whether the business is Recreation’ and ‘Specialised Design Service’ located within West Chelsea versus the refer- sectors. In a wider definition, design indus- ence area in Manhattan. Analyses 2 and 3 tries are also an art-related industries (also include the interaction terms, as products of called creative industries by, for example, year indicators and WCh (Year* WCh)to Florida, 2002; Currid, 2006), therefore, we allow the effect of WCh vary by year. made the Design category to include archi- Although the location variable WCh per se tectural design and the related engineering is time-invariant by definition, and should services, industrial and graphic design. be time-invariant in order not to violate the Nineteen other groups are also classified to proportionality assumption (see Singer and compose a full-list industry, and measured Willett, 1993 for details), the effect of the in the subsequent analyses along with arts locations may differ by year, reflecting local and cultural industries. In Appendix I, we circumstances. Since West Chelsea under- present a full categorisation, and corre- went multiple public interventions during sponding sample size of the industries. the observation period, making WCh act as time-variant. Therefore, in case the business closing and opening probabilities are calcu- Data-analytic plan lated based on the calendar year in Analysis Analysis 1: As an actively revitalising neigh- 2 and 3, the heterogeneous location effect bourhood, has West Chelsea been advanta- becomes systematic and causes a violation geous for late-arrived arts and cultural of the proportionality assumption. Thus, the industry to sustain their businesses, compared interactions were added to address this issue. with other neighbourhoods in Manhattan (the Twenty-two time-invariant industry Reference Area)? dichotomies, Industry, are included for two The event of interest is closing operation purposes: to control potentially distinctive from the current location, and the duration industry-specific business cycle by main is the number of surviving years counted effect, and to discern differential effect of from its opening. We fit the following mod- West Chelsea’s locations exerted on arts and els for the ith business establishment in the cultural, as well as other industries by inter- kth year, in a particular subsample of the actions with the question variable establishments that opened their doors after (Industry*WCh). We followed the NAICS 2000 (the ‘late-arrivers’). We first examine sector structure (U.S. Census, n.d.) for ini- the effect of the location in overall industry, tial categorisation, then conducted detailed without controlling Industry*WCh, then the subdivision, composition and omission term is added in the second analysis to dis- based on the site context. cern industry-specific effect of the location. Within the vector of Industry, the arts The same procedure applies to Analyses 2 and cultural industries (Gallery and Artists) and 3. and their interactions are the primary focus. ln (hik= )=aðÞYear + gðÞIndustry Museums, galleries within the ‘Arts, 1 hik Entertainment, and Recreation’ and art + d(Industry*WChi)+b1WChi ð1Þ dealers within the ‘Retail Trade’ sectors are combined to form a new category named log (hik= ) is log-odds of the probability 1 hik Gallery. The Artists category is created to of a business to close its operation, a(Year) include independent artists, photographers is the baseline hazard function of closing and their agencies or promoters, separated operation measured by age of establishment Yoon and Currid-Halkett 31

(a(Year)=a1YB1 + a2YB2 +a3YB3 + ...+ Analysis 3: Compared with the other neigh- a13YB13). WChi, the question predictor, pre- bourhoods in the reference area within sents the location effect of West Chelsea. Manhattan, did West Chelsea remain a com- g(Industry) is a function of 22 industries and paratively attractive location to open busi- d(Industry*WChi) is its interaction with the nesses for arts and cultural industries as question predictor. active revitalisation progressed in the neigh- bourhood? And, what other types of industries Analysis 2: As an actively revitalising neigh- constitute the ‘late-arrivers’? bourhood, has West Chelsea been advanta- In short, this analysis is the other side of geous for early-arrived arts and cultural Analysis 1. Whereas Analysis 1 focused on industry to sustain their businesses, compared closing businesses, this analysis focuses on with other neighbourhoods in Manhattan (the the opening businesses, demonstrating the Reference Area)? difference in probabilities of encountering a In this segment of analysis, the same new business establishment over an old one model of Analysis 1 is used with a minor in West Chelsea and the reference area in modification. As this research question Manhattan. The subsample of this analysis focuses on establishments that opened the includes business establishments that are doors before 2000 (the ‘early-arrivers’), the active as of 2011 and those that newly corresponding subsample is reversely selected opened and closed within the observation from Analysis 1; members opened the busi- period, and among them newly opened nesses before the observation period, i.e. left- establishments in each year of observation censored. The event in interest is closing period are investigated as opposed to older operation, and the duration is the number of ones of which their openings were prior to surviving years counted from 2000 not from the observation period. As noted in Analysis its opening year, while opening year of each 2, the baseline function concerns the calen- establishment is controlled. dar year (equation 3). The logic of this analysis is that the neigh- ln (hik= )=aðÞYear + sðÞYear*WCh bourhood revitalisation process is an exter- 1 hik i nal factor that disturbs the normal business + gðÞIndustry + d(Industry*WChi) lifetime cycle, which is assumed to occur + b WCh ð3Þ from 2000 and have lasted during the obser- 3 i vation period. Consequently, our objective In the three subsequent analytical results, the in this analysis is to estimate differing hazard question variable WCh (the location indicator probability of closing operation the busi- of West Chelsea) and its interactions deserve nesses establishments have faced from 2000 the most attention as they together depict the to 2012, depending on their locations. As location effect of West Chelsea, influencing mentioned before, we introduce Year*WCh, the probability of closing or opening of busi- to control heterogeneity of the location ness establishments. Positive fitted values of effects. Modified fitted model is below equa- the parameters associated with these predic- tion (2). tors signify larger probability of closing oper- ation compared with the reference location in ln (hik= )=aðÞYear + sðÞYear*WCh Model 1 and 2, and larger probability of new 1 hik i opening in Model 3. + rOpening Year + gðÞIndustry The series of Year variables capture the + d(Industry*WChi)+b2WChi ð2Þ general hazard profile of the events for all 32 Urban Studies 52(1) industries. The coefficients of Industry cov- whole, not particular industries (i.e. the arts, ariates display their relative deviation of professional services). probabilities to the events, from the hazard In model 1-a in Table 1, the fitted odds profile of omitted industries. Direct interpre- that a business establishment located in tation is made only on their interactions West Chelsea will close its operation are with WCh (Year*WCh and Industry*WCh), 0.708 times the fitted odds of the events for in order to keep the focus on the location their counterparts in the reference area effect of West Chelsea. (b = 0:346***, e0:346 = 0:708) (fitted The Year*WCh interactions reveal the dif- odds calculation will be omitted for simpli- ferential effect of the location in specific years city hereafter). More specifically, the ratio of made by local circumstance. Positive fitted the probability that a business establishment values of parameters associated with in West Chelsea will close its operation to Year*WCh mean particularly larger probabil- the probability that it will not is 0.708 times ity of closing business operation in the specific of those occurring in the reference area, years, resulting from adverse local incidents in meaning firms in West Chelsea are less likely Model 1 and 2, and larger probability of new to close their operation compared with their business openings in Model 3. Likewise, posi- counterparts (the full interpretation of odds- tive fitted value parameters associated with ratio will be omitted for simplicity here- Industry*WCh presents a larger risk of closing after). In industry-specific result in Model 1-b, business operation of the particular industries contrary to conventional wisdom about arts in West Chelsea, compared with the reference districts, at this stage in the development pro- area in Model 1 and 2, and a large chance of cess, art-related industries do not exhibit better observing new business opening in Model 3. success rate despite the colocation with other Among the 22 tested in this study, similar types of industry. Gallery*WCh, Artist*WCh and Desgin*WCh In Figure 3(1), to aid interpretation, we will be discussed in more depth. display the fitted probability of closing oper- ation (a) for overall industry and (b) for gal- lery industry, in each year of business in Results West Chelsea versus its counterparts in the In Table 1, we present estimates of log-odds reference area. West Chelsea pertains a of closing or opening businesses establish- lower hazard profile of closing operation in ment in each subsample will experience. general than the reference area, as evidenced Model 1-a and 1-b are of Analysis 1, first in Figure 3(1, a). without, then with the interaction term, Industry*WCh. Model 2-a and 2-b, and 3-a and 3-b explain the corresponding Analyses Analysis 2: Fitted odds of closing operation 2 and 3, respectively. for early-arrivers In contrast to the findings of Analysis 1, Analysis 1: Fitted odds of closing operation West Chelsea is a disadvantageous location to sustain businesses for early-arrivers over- for late-arrivers all, suggesting that extant gentrification The result of Analysis 1 suggests that late- hypotheses hold true for West Chelsea as arrivers located in West Chelsea have lower well. Early-arrived business establishments probability of closing operation compared in West Chelsea have larger probability of with their counterparts in the reference area. closing their operations compared with This location advantage affects industry as a those in the reference area. In the later years onadCurrid-Halkett and Yoon Table 1. Parameter estimates (with standard errors) and goodness-of-fit statistics for final fitted discrete-time hazard probability models observed between from 2000 to 2012 for three analyses. Analysis 1 (Model 1-a and 1-b): for businesses opened after 2000 describing business failure probabilityin a specific year in business, without industry*WCh interactions to see overall location effect (Model 1-a), and with industry*WCh to see industry-specific location effect (Model 1-b). Analysis 2 (Model 2-a and 2-b): for businesses opened before 2000 describing business failure probability in a specific year in observation period, without (Model 2-a) and with industry*WCh (Model 2-b). Analysis 3 (Model 3-a and 3-b): for business opened after 2000 describing business opening probability in a specific year in observation period, without (Model 3-a) and with industry*WCh (Model 3-b).

Analysis 1 Analysis 2 Analysis 3 Model 1-a Model 1-b Model 2-a Model 2-b Model 3-a Model 3-b

Variables 1st yr (yr 2000) 26.063*** 25.884*** 234.211*** 234.161*** 22.474*** 22.509*** (0.234) (0.250) (2.603) (3.098) (0.037) (0.050) 2nd yr (yr 2001) 24.952*** 24.910*** 234.208*** 234.106*** 22.678*** 22.706*** (0.148) (0.162) (2.601) (3.098) (0.040) (0.055) 3rd yr (yr 2002) 24.446*** 24.437*** 233.971*** 233.872*** 22.622*** 22.636*** (0.125) (0.135) (2.600) (3.097) (0.040) (0.055) 4th yr (yr 2003) 23.627*** 23.659*** 233.931*** 233.860*** 22.561*** 22.579*** (0.095) (0.105) (2.599) (3.097) (0.038) (0.055) 5th yr (yr 2004) 23.398*** 23.406*** 234.090*** 233.978*** 22.592*** 22.612*** (0.088) (0.100) (2.602) (3.097) (0.042) (0.057) 6th yr (yr 2005) 23.212*** 23.217*** 234.821*** 234.702*** 22.530*** 22.542*** (0.084) (0.097) (2.606) (3.098) (0.041) (0.058) 7th yr (yr 2006) 23.128*** 23.121*** 234.586*** 234.470*** 22.288*** 22.310*** (0.086) (0.099) (2.604) (3.098) (0.039) (0.055) 8th yr (yr 2007) 22.706*** 22.692*** 234.508*** 234.430*** 22.314*** 22.324*** (0.081) (0.092) (2.603) (3.098) (0.039) (0.057) 9th yr (yr 2008) 22.851*** 22.829*** 234.437*** 234.314*** 22.898*** 22.924*** (0.092) (0.102) (2.604) (3.097) (0.053) (0.072) 10th yr (yr 2009) 22.510*** 22.518*** 233.409*** 233.304*** 22.882*** 22.881*** (0.092) (0.102) (2.602) (3.096) (0.053) (0.073) 11th yr (yr 2010) 22.642*** 22.638*** 233.596*** 233.482*** 22.647*** 22.655*** (0.107) (0.118) (2.597) (3.096) (0.049) (0.068) 12th yr (yr 2011) 23.066*** 23.015*** 233.755*** 233.687*** 23.594*** 23.632*** (0.148) (0.161) (2.602) (3.096) (0.076) (0.104) 13th yr (yr 2012) 23.716*** 23.666*** 233.472*** 233.350*** 22

(0.267) (0.287) (2.603) (3.096) 33 (Continued) 34 Table 1. (Continued)

Analysis 1 Analysis 2 Analysis 3 Model 1-a Model 1-b Model 2-a Model 2-b Model 3-a Model 3-b

Yr 2000*WCh – – 20.221 20.185 0.079 0.177* (0.158) (0.200) (0.079) (0.106) Yr 2001* WCh – – 20.053 20.062 0.079 (0.193) (0.088) (0.117) Yr 2002* WCh – – 20.223 20.297 20.034 0.059 (0.149) (0.192) (0.088) (0.118) Yr 2003* WCh – – 0.184 0.117 0.060 (0.135) (0.178) (0.119) Yr 2004* WCh – – 20.112 0.201** 0.264** (0.197) (0.088) (0.118) Yr 2005* WCh – – 0.142 0.263 0.169* 0.245** (0.191) (0.234) (0.089) (0.119) Yr 2006* WCh – – 0.507*** 0.441** 20.054 0.068 (0.162) (0.209) (0.090) (0.120) Yr 2007* WCh – – 0.263 0.284 0.086 (0.171) (0.216) (0.124) Yr 2008* WCh – – 0.238 0.147 0.401*** 0.521*** (0.171) (0.218) (0.105) (0.140) Yr 2009* WCh – – 20.026 20.060 0.005 (0.169) (0.122) (0.162) Yr 2010* WCh – – 0.346*** 0.373** 0.203** 0.315** (0.129) (0.169) (0.097) (0.141) Yr 2011* WCh – – 0.396*** 0.430** 0.373** 0.606*** (0.139) (0.181) (0.145) (0.193) Yr 2012* WCh – – 0.216 0.225 – – (0.136) (0.176) 52(1) Studies Urban Gallery 20.039 20.434 0.115 21.120*** 0.144*** 20.349*** (0.145) (0.286) (0.088) (0.248) (0.053) (0.112) Artist 20.168 20.280 0.273*** 20.672*** 20.191*** 20.329*** (0.156) (0.249) (0.080) (0.182) (0.057) (0.108) Design 20.225* 20.427* 20.713*** 20.824*** 20.005 20.244*** (0.120) (0.226) (0.091) (0.177) (0.043) (0.094) (Continued) onadCurrid-Halkett and Yoon Table 1. (Continued)

Analysis 1 Analysis 2 Analysis 3 Model 1-a Model 1-b Model 2-a Model 2-b Model 3-a Model 3-b

Const. 0.219 0.183 20.005 20.303** 20.122** 20.117 (0.140) (0.179) (0.090) (0.152) (0.059) (0.091) Manuf. 0.161 0.140 0.218*** 20.048 20.519*** 20.433*** (0.121) (0.148) (0.060) (0.094) (0.051) (0.077) Auto. 1.535*** 0.543 0.611*** 0.816** 20.317*** 20.017 (0.204) (0.600) (0.130) (0.319) (0.110) (0.314) Whs. 0.114 0.093 0.250*** 0.229*** 20.375*** 20.339*** (0.115) (0.143) (0.062) (0.086) (0.048) (0.073) Retail 0.129 0.136 20.062 20.107 0.033 0.059 (0.081) (0.099) (0.053) (0.072) (0.032) (0.048) Info. 0.003 0.025 20.216*** 20.224** 0.040 0.038 (0.112) (0.135) (0.080) (0.111) (0.045) (0.068) FIRE 20.280*** 20.239* 20.515*** 20.522*** 20.188*** 20.162*** (0.104) (0.124) (0.070) (0.098) (0.038) (0.058) B.Support 20.068 20.039 20.184** 20.112 20.184*** 20.147* (0.156) (0.181) (0.093) (0.123) (0.060) (0.086) Tech. 20.344 20.214 0.107 0.032 20.075 0.065 (0.338) (0.405) (0.221) (0.304) (0.119) (0.169) Mng. 20.258 20.261 0.274 20.231 0.667*** 0.686*** (0.447) (0.513) (0.462) (0.488) (0.165) (0.226) Legal 20.403*** 20.433*** 20.552*** 20.617*** 20.202*** 20.186*** (0.135) (0.156) (0.089) (0.115) (0.046) (0.065) Admin. 0.158 0.149 0.082 20.027 20.320*** 20.298*** (0.123) (0.140) (0.071) (0.094) (0.051) (0.073) Edu. 20.389 20.254 20.458*** 20.381* 20.501*** 20.460*** (0.287) (0.329) (0.169) (0.217) (0.098) (0.148) Health 20.903*** 20.943*** 20.886*** 20.859*** 20.272*** 20.263*** (0.179) (0.220) (0.111) (0.147) (0.050) (0.075) Etm. 0.581*** 0.379 20.447*** 20.279 20.108 20.363 (0.165) (0.395) (0.138) (0.297) (0.076) (0.221) Lodging 20.511 20.385 20.889** 21.247** 20.151 20.377 (0.516) (0.720) (0.352) (0.583) (0.140) (0.251) 35 (Continued) 36 Table 1. (Continued)

Analysis 1 Analysis 2 Analysis 3 Model 1-a Model 1-b Model 2-a Model 2-b Model 3-a Model 3-b

Restr. 0.340*** 0.262** 0.211*** 0.317*** 0.288*** 0.356*** (0.098) (0.128) (0.074) (0.108) (0.041) (0.065) P.Svc. 0.100 0.102 0.050 0.013 20.031 20.030 (0.155) (0.195) (0.097) (0.144) (0.063) (0.098) Pb.Svc. 20.278 20.227 20.954*** 21.096*** 20.532*** 20.503*** (0.259) (0.289) (0.182) (0.245) (0.088) (0.126) Gallery* WCh – 0.162 – 1.658*** – 0.553*** (0.355) (0.284) (0.143) Artist* WCh – 0.477 – 1.567*** – 0.312** (0.366) (0.227) (0.157) Design* WCh – 0.540* – 0.280 – 0.404*** (0.324) (0.250) (0.131) Const. * WCh – 0.385 – 0.765*** – 0.052 (0.402) (0.245) (0.195) Manuf. * WCh – 0.388 – 0.764*** – 20.363** (0.389) (0.169) (0.183) Auto. * WCh – 1.385** – 20.104 – 20.324 (0.668) (0.363) (0.360) Whs. * WCh – 20.022 – 0.230 – 20.227 (0.381) (0.191) (0.180) Retail* WCh – 0.365 – 20.068 – 20.004 (0.251) (0.154) (0.104) Info.* WCh – 0.019 – 20.016 – 0.100 (0.371) (0.268) (0.158) FIRE* WCh – 20.118 – 20.112 – 20.042 (0.377) (0.229) (0.130) 52(1) Studies Urban B.Support* WCh – 20.449 – 0.085 – 20.016 (1.036) (0.485) (0.381) Tech. * WCh – – 1.375 – 0.770 (1.111) (1.112) Mng.* WCh – – –

(Continued) onadCurrid-Halkett and Yoon Table 1. (Continued)

Analysis 1 Analysis 2 Analysis 3 Model 1-a Model 1-b Model 2-a Model 2-b Model 3-a Model 3-b

Legal* WCh – 1.305* – 0.367 – 0.126 (0.769) (0.738) (0.405) Admin. * WCh – 0.422 – 0.716*** – 20.062 (0.631) (0.251) (0.277) Edu. * WCh – – 21.213 – 20.122 (1.034) (0.350) Health* WCh – 0.517 – 20.626 – 0.034 (0.580) (0.534) (0.182) Etm. * WCh – 0.497 – 20.074 – 0.340 (0.476) (0.376) (0.256) Lodging* WCh – 20.106 – 1.142 – 0.705** (1.253) (0.936) (0.303) Restr. * WCh – 0.513* – 20.296 – 20.154 (0.280) (0.218) (0.127) P.Svc. * WCh – 0.188 – 0.111 – 0.081 (0.466) (0.273) (0.202) Pb.Svc. * WCh – – 1.586*** – 20.333 (0.504) (0.480) Open_Year – – 0.016*** 0.015*** –– (0.001) (0.002) WCh 20.346*** 20.589*** 0.238*** 0.110** (0.070) (0.196) (0.073) (0.052) N 51,599 51,610 89,907 89,890 106,374 106,379 d.f. 36 54 47 69 46 69 F-test or 22LL 525 12,079 985 23,726 1,738 48,203

Note: Standard errors in parentheses ***p \ 0.01, **p \ 0.05, *p \ 0.1, – implies variables that have not been included, and crossed variables are dropped because of multicollinearity. 37 38 Urban Studies 52(1)

Figure 3. (1) Fitted probability of business closure for overall industries (a) and fitted probability of business closure for galleries (b) in each year in business, for establishments opened after 2000 (late- arrivers), separately plotted by locations (WCh versus the reference area), ((a) is plotted from Model 1-a; Table 1, (b) is plotted with subsamples of gallery industries). (2) Fitted probability of business closure in each year in observation, for overall establishments opened before 2000 (early-arrivers), separately plotted by locations (WCh versus the reference area), ((a) is plotted from Model 2-a; Table 1, (b) is plotted with subsamples of gallery industries). of the observation period as West Chelsea portion measured in the reference area in becomes more fully developed, this adverse Manhattan. This confirms the prevailing location effect becomes exacerbated. Our expectation that early-arrived art industries results are specifically germane with regard in West Chelsea experienced more severe to art and cultural industries. Early-arrived displacement than those in other parts of art-related and cultural industries experi- Manhattan when the area undergoes active enced particularly low survival rates in West revitalisation, upholding previous hypothesis Chelsea. Within these specific industries, a on the arts and gentrification (Currid, 2007; disproportionately larger portion of them Lloyd, 2006; Zukin, 1989, among others). failed to keep the business in their current Along with these, construction, manufactur- locations in each year, compared with the ing, minor business administration, public Yoon and Currid-Halkett 39 affairs, also experienced a particularly hard gap in two different locations observed in time. 2006, 2010 and 2011, although we cannot In Model 2-a, the fitted odds of closing find a causal link in this analytical structure. operation are 1.269 times larger for overall This hardship is only to the early-arrived businesses in West Chelsea, compared with businesses of West Chelsea and the newer those in the reference area in Manhattan ones established after 2000 are not at greater (b = 0:238***). In the years of 2006, 2010 risk of closing their doors than their counter- and 2011, the positive values of the interac- parts in the reference area, as presented in tions Year* WCh enlarge the difference of the Analysis 1. It is probable that the fron- probabilities between those two locations tier businesses might not have expected an (Model 2-b). In Figure 3(2), we visualise the imminent upward transformation of the fitted hazard profiles of closing operation by West Chelsea to a new commercial/residen- location; WCh versus the reference area, (a) tial district and tourist attraction, when they for overall and (b) for gallery industry. West moved in. Thus, in the current high-end Chelsea pertains the higher hazard profile commercial real estate market, the viability than the reference area, and in 2006, 2010 of those early-arrivers would be less robust and 2011, the gap becomes even larger. than for followers, and susceptible to ‘price- Industry-specific results in Model 2-b outs’. identify that Gallery, Artists, Construction, Manuf., Admin., Pb.Svc. had undergone dis- proportionately higher business closure in Analysis 3 : Fitted odds of opening West Chelsea. West Chelsea imposes partic- businesses (late-arrivers) ularly larger hardship on Gallery,inthat West Chelsea accommodates comparatively their survival chance is lower than their higher frequency of business openings than counterparts by the largest margin, evi- those of the reference area in Manhattan, denced by 5.249 times larger fitted odds of south of the Central Park. The probability closing operation (Gallery*WCh; of encountering a new business is high b = 1:658***) at a statistically significant throughout the observation period, with level of less than 5%. Artists in West greater intensity in years of 2004, 2005, Chelsea are also in a comparatively vulnera- 2008, 2010 and 2011. New openings of ble position, exhibiting 4.792 times of the design, gallery industry and artists’ enter- fitted odds of business closure compared prises are visible in West Chelsea, compared with their counterparts (Artists*WCh; with the level observed in the reference area. b = 1:567***): This outcome can be interpreted as a sign of As discussed, one of the factors that neighbourhood-level industrial clustering or explains the higher vulnerability of early- localisation in the post-2000 period. These arrived business establishments in West industrial concentrations reflect the conven- Chelsea is the upwardly shifting neighbour- tional observation that West Chelsea has hood characteristics and ongoing economic emerged as New York City’s new arts dis- development of the area, partly stimulated trict, and with it come the accoutrements of by aforementioned planning interventions the ‘creative class’ (Currid, 2010; Molotch and public projects. The announcement of and Treskon, 2009). Contrary to the con- rezoning in 2005 and completion of the High cerns voiced in the media that West Chelsea Line in 2009 and 2011, which may have is losing its unique characteristics as an arts facilitated the gentrification process, are and cultural enclave, our empirical work roughly aligned with the increased hazard here indicates that in reality the 40 Urban Studies 52(1)

Figure 4. (a) Fitted probability of business opening in each calendar year from 2000 to 2011, for overall establishments, separately plotted by locations (WCh versus the reference area), (plotted from Model 3-a; Table 1). (b) (c) (d) Fitted probability of business opening in each calendar year from 2000 to 2011, for Gallery, Design and Lodging establishments separately plotted by locations (WCh versus the reference area), (plotted from Model 3-b; Table 1 with subsamples of corresponding industries). neighbourhood remains a compelling place of West Chelsea deviates from the propor- for new gallery, design and art enterprises. tional trajectory in the year 2004, 2005, Another visible influx is found in lodging 2008, 2010 and 2011, displaying even stron- industry (i.e. hotels) foreshadowing the com- ger probability of firm openings than the modification of the neighbourhood. As one expected solely by the main effect. expected, less manufacturing enterprises Figure 4(a) depicts new business opening choose West Chelsea for their new business probability profiles by location; West location. Chelsea exhibits the higher proportion of For the overall industries, the fitted odds newly opening to existing businesses com- of observing a new business over an old one pared with the reference area in general, and is higher by 1.116 times (b = 0:110***) of the gap increases more in the later years, the corresponding odds in the reference area presented with interaction terms. (Model 3-a). By the positive values of inter- According to the industry-specific results action terms, Year* WCh, the location effect in Model 3-b, a higher proportion of Design, Yoon and Currid-Halkett 41

Gallery, Artists, Lodging industries open example, the later galleries and art firms their businesses in West Chelsea over the ref- may possess greater business acumen (per- erence area, while new openings in manufac- haps locating after the neighbourhood trans- turing are rare in West Chelsea. The fitted formed is a sign of such skill). The later odds of observing a newly opened gallery in galleries may also have more economic West Chelsea during the observation years resources (higher rents in the later years are 1.738 times greater than the fitted odds would suggest as much). Early galleries and of doing so in the reference area art firms may have been motivated by the (Gallery*WCh; b = 0:553***). To a lesser abandoned space and cheaper rents, as degree, independent art enterprises and SoHo would have been the premier arts dis- design related industry also have higher trict and also more expensive. Regardless of opening tendency in West Chelsea, suggest- the myriad motivations and explanations, it ing the neighbourhood’s increasing impor- is clear that the arts and neighbourhood tance as an arts district. Lodging industry development model as put forth by Zukin also joins this trend, in that the fitted odds (1989) and more generally discussed by of observing new lodging venues over old other arts scholars (Currid, 2007; Lloyd, ones in West Chelsea is 2.024 times higher 2006) holds true. than that of the reference area (Lodging*WCh; b = 0:705***). Figure 4(b)– (d) illustrates these statistical outcomes. Limitation The analytical results on gallery industry One of the limitations of this study is the and artists enterprises, together with their selection of the watershed year. Year 2000 is higher probability of closing operation used to divide early-arrivers and late-arrivers revealed in the Analysis 2, suggests a fast in the current study, but could have been turnover rate in West Chelsea, compared established differently. Ideally, the year 2000 with the reference area. As of the plausible should split the businesses into two distinc- turning point of year 2000, early-arrived gal- tive groups that have contrasting economic leries and artists businesses in West Chelsea positions. Although a series of public inter- are more prone to business closure than ventions facilitated the revitalisation process those in the reference area, whereas a pro- of West Chelsea from 2000 onwards, the eco- portionally large number of new galleries nomic restructuring takes much longer than and art-related businesses continued to open a year. Therefore, any year within a plausible in West Chelsea after 2000, and no longer range from 2000 could be chosen for the have the systematically higher tendency of watershed year, without harming the logic of closing operation, as evidenced in Analysis this analytical design. In any case, the mag- 1. nitudes of the estimates would be affected This outcome suggests that early pio- depending on the selection at a minor level, neers, which may have been less affluent or while the direction of the estimates would economically successful to begin with, were remain the same as the current study. pushed out, but that later galleries, perhaps Another limitation may be in the estab- lured by the neighbourhood’s new reputa- lishment of the boundary of West Chelsea in tion as an arts district, were more resilient our analysis. We attempted to define the site and successful. We caution that there are of investigation to accurately capture the full many other characteristics that may explain extent of the location effect of West Chelsea. the lesser success of the pioneers and the In order to do this, we based our analysis on greater success of the late adopters. For reasonable assumptions and sensitivity 42 Urban Studies 52(1) analysis. However, since the precise extent As witnessed in SoHo, did art and cul- of the location impact of West Chelsea is tural industries lose their position in West not fully known, we could have delineated Chelsea? At least for the early pioneer gal- the boundary of West Chelsea around the leries and art-related firms, the West Chelsea core West Chelsea area alternatively, with outcome mirrors that of SoHo. Although a differing assumptions. Likewise, the magni- causal link cannot be validated in this study, tudes of the estimates would be affected by we can attribute this to the progress of revi- the change, while the direction of the esti- talisation. Since Chelsea’s 1999 rezoning, the mates would be the same. observation period saw West Chelsea rezon- ing and the High Line completion in 2005, 2009 and 2011, respectively. Although these Conclusion changes may have increased the desirability Our work attempts to unpack the relationship of the neighbourhood, it may have adversely between artistic industries and economic devel- affected the pre-existing businesses in their opment, based on empirical analyses on the survival as the area became more appealing business survival in West Chelsea between to a wide variety of businesses and people. 2000 and 2012, the most active postindustrial This reasoning also fits to the analytical revitalisation period for the area. Whilst we do result of the businesses that opened in the not claim a causal link between the arts and later phase of the revitalisation. We find that development, we have gained a greater under- for overall industries opened in the more standing of how the arts relate to the neigh- rapidly developing period (post-2000), West bourhood development and gentrification Chelsea is a better environment for survival cycle. Our work upholds much of the previ- for businesses. Indeed West Chelsea is an ous, primarily qualitative work done in this attractive neighbourhood for new business area. We find that the early arts and cultural opening for particular types of industries, industries (those that arrived prior to 2000) do compared with the reference area in not fare well as the neighbourhood develops, Manhattan. The influx of galleries and artist particularly as we compare West Chelsea with firms are visibly higher, corroborating our Manhattan reference area. extant reports in the scholarship and media We find that for overall industries opened detailing West Chelsea’s ascent as New before 2000, West Chelsea is not a favour- York City’s new art district. Higher propor- able environment for survival. Early-estab- tions of design industries also open their lished firms have failed to sustain themselves doors in West Chelsea, signifying West in their original locations compared with Chelsea has attracted other types of creative those in other parts of Manhattan, as industries, suggesting some of the creative observed in the lower survival probability. class dynamics put forth by Florida (2002). Arts and cultural industries are particularly This analytical result fairly documents that hard hit. The early-arrived museums, art the art/cultural industry which initiated galleries, artists’ working spaces in West postindustrial restructuring of West Chelsea Chelsea are more susceptible to closing oper- still maintains a significant concentration in ation than their counterparts in the reference the neighbourhood, even as it diversifies. area, setting aside New York City’s well- This contrast between early-arrived and documented manufacturing firm loss from late-arrived arts and cultural industries the 1960s. Administrative and supportive implies that the nature of the art scene in services, non-profit and public services are West Chelsea may have changed during the in the same position. course of the neighbourhood revitalisation Yoon and Currid-Halkett 43

(as it did in Soho). Although the characteris- popular spots in New York (PHA, 2005). tics of galleries in their rising and falling Residential development has increased dra- phases cannot be deduced from this analysis, matically (Gregor, 2010; Russell, 2011), Schuetz et al.’s (2011) study suggests that and developers have sought zoning var- higher-tier galleries tended to survive longer iances to acquire even more FAR to survive than smaller or less economically robust gal- under intensified property development leries. As such, upward succession (and more competition (Berman, 2013). This suggests high-tier galleries) may also have occurred in that localisation economies are at play West Chelsea. An interview conducted by within the gallery and arts industries and Kazakina (2013) gave a clue that mid-sized urbanisation economies are beneficial more galleries were struggling from the raised rent, broadly. while the ‘big players’2 in the industry have As this research and extant literature expanded their presence in West Chelsea. argues, the arts and cultural industries, in The cause of the continuing agglomera- general, are critical in the early stage of the tion of arts and cultural industries in West neighbourhood revitalisation process. They Chelsea cannot be fully explained through have also emerged as significant players in the scope of this research. However, it can the wider economic dynamics within cities be reasonably understood as ‘path-depen- and regions (Currid, 2006; Scott, 2000, 2005; dency’, whereby specialised services and Stolarick and Currid-Halkett, 2013, among knowledge-based industry cluster in an itera- others). This is not limited to US cities, but tive fashion to share the collective social, observed ubiquitously in world cities intellectual and operational resources (Economic Survey of Singapore, 2003; (Boschma and Lambooy, 1999; Currid, Fujita and Child Hill, 2004; Yusuf and 2007; Scott, 2006). Arts and cultural indus- Nabeshima, 2005). tries in West Chelsea may benefit from the In order to maintain the specific charac- agglomeration economies. However, exces- teristics of the place and maximise the syner- sive competition between similar types of gic effect of the agglomeration of creative, business inevitably accompanies. Amid the arts and cultural industries, balanced policy heated competition for space and customers, intervention is necessary (Mommaas, 2004). businesses with increased means can acquire As this research suggests, the double goals space more easily than others. of West Chelsea rezoning to boost the devel- We find that this spillover effect applies opment momentum and to preserve the orig- to overall businesses and firms, presented as inal nature of the art scene is inevitably self- low business closure probability in the analy- conflicting, and requires greater effort to sis. Given the movement of New York City’s find the balance between those two. The cur- art district to West Chelsea, the neighbour- rent analysis supports the partial success of hood gained more visits from customers the policy on the second count, in that West which undoubtedly supported the economic Chelsea maintains an advantage in arts- viability of in situ businesses. The fast- related industries and galleries. However, growing hotel industry in West Chelsea evi- these policies and rezoning efforts could not denced in this study, and widely reported prevent the ‘price-out’ of the earlier retail expansion including higher-end restau- pioneers. rants and fashion boutique shops (Moss, Recall however that SoHo’s movement 2012a; PHA, 2005), and the increase in ame- from nascent arts community to commercial nities – the High Line, the Hudson River takeover spanned over 25 years. Whilst the Park – have made West Chelsea one of the DCP rezoning for artists’ live-work space in 44 Urban Studies 52(1)

1971, and the galleries moved in through the Beauregard RA (2009) Urban population loss in 1970s and 1980s, it was only in the 1990s historical perspective: United States, 1820- when Soho’s galleries started closing or mov- 2000. Environment and planning A 41(3): 514. ing out (Currid, 2010; Zukin, 1989). Thus, Berman A (2013) High Line’s Open Views Soon the West Chelsea arts story continues to To Disappear.17January.Off the Grid. Avail- unfold. Although a longer time must pass able at: http://gvshp.org/blog/2013/01/17/high- lines-open-views-soon-to-disappear/. before we conclude on the success of the Best Manhattan Offices (2013) Manhattan Com- planning intervention, already our research mercial Market Overview. Best Manhattan would indicate that a more aggressive and Offices. Available at: http://bestmanhattanof- comprehensive regulatory framework could fices.com/manhattan-commercial-submarket- have been instrumental in promoting such overview/. industries in this context. Bondi L (1999) Gender, class, and gentrification: Enriching the debate. Environment and Plan- ning D 17: 261–282. Funding Boschma RA and Lambooy JG (1999) Evolu- This research received no specific grant from any tionary economics and economic geography. funding agency in the public, commercial, or not- Journal of Evolutionary Economics 9(4): for-profit sectors. 411–429. Brooks D (2010) Bobos in paradise: The new upper class and how they got there. New York: Simon Notes and Schuster. 1. Unlike SoHo, West Chelsea began to attract Brozan N (2004) Homes start to invade a gallery art dealers first and artists later. Even before neighborhood. The New York Times,21 the art dealers made their way, however, there November. has been a unique art scene created by photo- Bruni F (2008) Genre-bending hangout takes its graphers and performing artists (Bruni, 2008; final bows. The New York Times,21May. Pincus, 1997). We do not claim that there is a Available at: http://www.nytimes.com/2008/05/ strong connection between the art works cre- 21/dining/21florent.html?pagewanted=all&_r=0. ated and consumed in West Chelsea. Cameron S and Coaffee J (2005) Art, gentrifica- 2. For example, David Zwirner, Hauser and tion and regeneration–from artist as pioneer to Wirth, Larry Gagosian, Pace and Barbara public arts. European Journal of Housing Pol- Gladstone expanded their galleries or opened icy 5(1): 39–58. another branch in West Chelsea (Kazakina, Center for an Urban Future (2011) Growth by 2013). Design: The Powerful Impact & Untapped Potential of NYC’s Architecture & Design Sec- tor. New York: Center for an Urban Future. References Clay PL (1979) Neighborhood renewal. Lexington, Aalen OO, Borgan Ø, Gjessing HK, et al. (2008) MA: Lexington Books. Survival and Event History Analysis: A Process Costar Group (n.d.) Available at: http://www. Point of View. Springer. costar.com/about/SubmarketMaps.aspx. Abu-Lughod JL (1994) From Urban Village to Cox DR (1972) Regression models and life-tables. East Village: The Battle for New York’s Lower Journal of the Royal Statistical Society. Series East Side. Cambridge, MA: Blackwell. B (Methodological) 34(2): 187–220. Allison PD (1984) Event History Analysis: Regres- Curran W (2007) ‘From the frying pan to the sion for Longitudinal Event Data. SAGE Publi- oven’: Gentrification and the experience of cations, Incorporated. industrial displacement in Williamsburg, Beard R and Berlowitz L (1993) Greenwich Vil- Brooklyn. Urban Studies 44(8): 1427–1440. lage: Culture and Counterculture. Rutgers Uni- Currid E (2006) New York as a global creative versity Press. hub: A competitive analysis of four theories Yoon and Currid-Halkett 45

on world cities. Economic Development Quar- in Chicago, 1978–1987. Journal of the Ameri- terly 20(4): 330–350. can Planning Association 54(3): 279–290. Currid E (2007) How art and culture happen in Gregor A (2010) As a park runs above. The New New York: Implications for urban economic York Times, 10 August. development. Journal of the American Planning Holusha J (1997) Commercial property/West Association 73(4): 454–467. Chelsea; ex-garages attracting art galleries Currid E (2010) Symposium introduction – Art and from SoHo. The New York Times,12 economic development: New directions for the October. growth of cities and regions. Journal of Planning Hudson JR (1984) SoHo A study of residential Education and Research 29(3): 257–261. invasion of a commercial and industrial area. David J and Hammond R (2011) High Line: The Urban Affairs Review 20(1): 46–63. Inside Story of New York City’s Park in the Hudson JR (1987) The Unanticipated City: Loft Sky (First Edition). New York: Farrar, Straus Conversions in Lower Manhattan. University and Giroux. of Massachusetts Press. Dawid I (2012) The downside of the High Line.Pla- Hutton TA (2009) Trajectories of the new econ- netizen. Available at: http://www.planetizen. omy: Regeneration and dislocation in the inner com/node/58191. city. Urban Studies 46(5–6): 987–1001. Drennan M (1991) The decline and rise of the Hyra DS (2008) The New Urban Renewal: The New York economy. In: Mollenkopf JH and Economic Transformation of Harlem and Bron- Castells M (eds) Dual City: Restructuring New zeville. University of Chicago Press. York. Russell Sage Foundation, pp. 25–42. Kazakina K (2013) Desperate art galleries give up Economic Survey of Singapore (2003) Economic as Chelsea rents double. Bloomberg.Com,20 Contributions of Singapore’s Creative Industries. February. Singapore: Economic Survey of Singapore. Kerstein R (1990) Stage models of gentrification Fainstein SS (1999) The City Builders: Property an examination. Urban Affairs Review 25(4): Politics and Planning in New York and London. 620–639. Oxford: Wiley-Blackwell. Lees L (2000) A reappraisal of gentrification: Fainstein SS (2007) Tourism and the commodifica- Towards a ‘Geography of Gentrification’. tion of urban culture. The Urban Reinventors Progress in Human Geography 24(3): 389–408. Issue 2, 7 November. Ley D (1996) The new middle class and the remak- Fainstein SS and Judd D (1999) Cities as places to ing of the central city. New York: Oxford Uni- play. In: Fainstein SS and Judd D (eds) The versity Press. Tourist City. Yale University Press, pp. 261–272. Ley D (2003) Artists, aestheticisation and the Florida R (2002) Bohemia and economic geogra- field of gentrification. Urban Studies 40(12): phy. Journal of Economic Geography 2(1): 55–71. 2527–2544. Florida RL (2005) Cities and the Creative Class. Lloyd RD (2006) Neo-bohemia: Art and Com- Psychology Press. merce in the Postindustrial City. New York: Fujita K and Child Hill R (2004) Innovative Routledge. Tokyo. World Bank Policy Research Paper McGeehan P (2011) The high line isn’t just a sight No. 3507, 2005, presented at the World Bank to see; It’s also an economic dynamo. The New Workshop on ‘Creative Industries in East York Times, 5 June. Asia’. Bangkok, Thailand. Available at: http:// Markusen A and Schrock G (2006) The artistic papers.ssrn.com/sol3/papers.cfm?abstract_id= dividend: Urban artistic specialization and 660088 economic development implications. Urban Gale DE (1979) Middle class resettlement in older Studies 43(10): 1661–1686. urban neighborhoods: The evidence and the Markusen AR, Schrock G and Cameron M implications. Journal of the American Planning (2004) The Artistic Dividend Revisited. Project Association 45(3): 293–304. on Regional and Industrial Economics, Hum- Giloth R and Betancur J (1988) Where downtown phrey Institute of Public Affairs, University of meets neighborhood: Industrial displacement Minnesota. 46 Urban Studies 52(1)

Mele C (2000) Selling the Lower East Side: Cul- Ouroussoff N (2006) On the High Line, solitude ture, Real Estate, and Resistance in New York is pretty crowded. The New York Times,24 City. University of Minnesota Press. December. Molotch H and Treskon M (2009) Changing art: Philip Habib & Associates (2005) SpecialWestChel- SoHo, Chelsea and the dynamic geography of sea District Rezoning and High Line Open Space galleries in New York City. International Jour- Environmental Impact Statement.NewYork: nal of Urban and Regional Research 33(2): New York City Department of City Planning. 517–541. Pincus M (1997). Meat the Neighbors. New York Mommaas H (2004) Cultural clusters and the Magazine, 3 March, pp. 26–35. post-industrial city: Towards the remapping of Reis (n.d.) Market reports – Neighborhoods/ urban cultural policy. Urban Studies 41(3): Submarket sections. Available at: https:// 507–532. www.reis.com/market-reports-single-metro- Morgan R (2012) Where art studio bloomed. office/new-york/new-york. Available at: http://online.wsj.com/news/articles/ Rose D (1984) Rethinking gentrification: Beyond SB10000872396390443687504577565512621637478 the uneven development of Marxist urban the- (accessed 2 August). ory. Environment and Planning D: Society and Moss J (2012a) Disney World on the Hudson. Space 2(1): 47–74. The New York Times, 21 August. Rose D (1996) Economic restructuring and the Moss J (2012b) High Line Op-Ed response. diversification of gentrification in the 1980s: A Available at: http://vanishingnewyork. view from a marginal metropolis. Toronto: Uni- blogspot.kr/2012/08/high-line-op-ed-response.html versity of Toronto Press. (accessed 30 August). Russell J (2011) Million-dollar condos bulge, sway NYbits (n.d.) Chelsea. Available at: http:// above Manhattan’s High Line. Bloomberg.com. www.nybits.com/manhattan/chelsea/. Available at: http://www.bloomberg.com/news/ New York City, Department of City Planning 2011-05-03/million-dollar-condos-bulge-sway- (1996) The Chelsea Plan. New York: City of above-manhattan-s-high-line.html New York, Department of City Planning. Sassen S (1994) Cities in a World Economy (Vol. New York City, Department of City Planning 3). Thousand Oaks, CA: Pine Forge Press. (1999) Chelsea rezoning. Available at: http:// Schuetz J, Currid-Halkett E and Green R (2011) www.nyc.gov/html/dcp/html/zone/wn111599. Is the art market more bourgeois than bohe- shtml. mian? Available at: SSRN 1896514 (http:// New York City, Department of City Planning papers.ssrn.com/sol3/papers.cfm?abstract_id= (2001) Far West Midtown A Framework for 1896514). Development. New York: City of New York, Scott AJ (2000) The Cultural Economy of Cities: Department of City Planning. Essays on the Geography of Image-producing New York City, Department of City Planning Industries. Sage. (2006a) The Zoning Handbook.NewYork:City Scott AJ (2005) On Hollywood: The Place, The of New York, Department of City Planning. Industry. Princeton, NJ: Princeton University New York City, Department of City Planning Press. (2006b) Innovative New York zoning district Scott AJ (2006) Entrepreneurship, innovation and receives national planning award. Available industrial development: Geography and the at: http://www.nyc.gov/html/dcp/html/about/ creative field revisited. Small Business Econom- pr011206.shtml. ics 26(1): 1–24. New York City Landmarks Preservation Com- Simpson CR (1981) SoHo: The Artist in the City. mission (2008) West Chelsea Historic District Chicago, IL: University of Chicago Press. Designation Report. New York: New York Singer JD and Willett JB (1991) Modeling the City Landmarks Preservation Commission. days of our lives: Using survival analysis when New York City Planning Commission (1981) designing and analyzing longitudinal studies Lofts: Balancing the Equities. New York: New of duration and the timing of events. Psycho- York City Planning Commission. logical Bulletin 110(2): 268–290. Yoon and Currid-Halkett 47

Singer JD and Willett JB (1993) It’s about time: bourhood renewal processes in Brussels and Using discrete-time survival analysis to study Montreal. Urban Studies 40(12): 2451–2468. duration and the timing of events. Journal of Watt P (2005) Housing histories and fragmented Educational and Behavioral Statistics 18(2): middle class careers: The case of marginal pro- 155–195. fessionals in London council housing. Housing Smith N (2002) New globalism, new urbanism: Studies 20(3): 359–381. Gentrification as global urban strategy. Anti- Whyte WH (1956) The Organization Man. Simon pode 34(3): 427–450. & Schuster. Smith N (2012) The New Urban Frontier: Gentrifi- Wilson C (2005) Turning the High Line into . cation and the Revanchist City. Routledge. the high life. The New York Times,18 Smith N and DeFilippis J (1999) The reassertion December. of economics: 1990s gentrification in the lower Yamaguchi K (1991) Event History Analysis. east side. International Journal of Urban and Newbury Park, CA: Sage Publications. Regional Research 23(4): 638–653. Yusuf S and Nabeshima K (2005) Creative Soffer JM (2010) Ed Koch and the Rebuilding of industries in east Asia. Cities 22(2): New York City. Columbia University Press. 109–122. Stolarick K and Currid-Halkett E (2013) Crea- Zukin S (1989) Loft Living: Culture and Capital in tivity and the crisis: The impact of creative Urban Change. Rutgers University Press. workers on regional unemployment. Cities Zukin S (2010) Naked City: The Death and Life 33: 5–14. of Authentic Urban Places. Oxford University United States Census Bureau (n.d.) North Ameri- Press. can Industry Classification System. Available Zukin S and Braslow L (2011) The life cycle of at: http://www.census.gov/cgi-bin/sssd/naics/ New York’s creative districts: Reflections on naicsrch?chart=2012 (accessed 21 March 2013). the unanticipated consequences of unplanned Van Criekingen M and Decroly J (2003) Revi- cultural zones. City, Culture and Society 2(3): siting the diversity of gentrification: Neigh- 31–140. 48 Appendix I Industry category with NAICS codes, and number of businesses in the population data and sample, by locations (study area and WCh).

List (Variable name) NAICS codes Description Number in the population data Number in the sample Study area WCh Study area WCh

Gallery/Museum 712110~712190 Museums, Historic Sites 2093 628 408 540 (Gallery) 453920 and Similar institution, Art Dealer, Gallery Artists 711110~711190 Performing Arts and 4469 327 511 274 (Artist) 711310~711510 Related Industries, 541921 Independent Artists and Agencies, Photographers Architectural design and 541310~541380 16,356 302 855 225 related service, graphic design (Design) Construction 236115~238990 6249 173 554 128 (Const.) Manufacturing 311111~339999 13,948 429 1049 277 (Manuf.) Automotive Services 811111~811198 Originally from ‘‘Services’’ 766 221 297 167 (Auto.) sector Wholesale Trade (Whs.) 423110~425120 12,154 219 1002 173 Retail Trade (Retail) 441110~453910 Subsector ‘‘Art Dealer’’ 42,846 994 2154 686 453920 (453920) is moved to Gallery/Museum Information (Info.) 511110~519190 16,268 322 787 135 Finance, Insurance, Real 521110~533110 21,831 342 1415 242 Estate, Rental Leasing (FIRE) 52(1) Studies Urban Accounting, Consulting, 541211~541219 12,434 229 650 171 Tax Preparation, 541611~541690 Advertising, Public 541810~541890 Relation (Business Support) (Continued) Appendix I. (Continued) Currid-Halkett and Yoon

List (Variable name) NAICS codes Description Number in the population data Number in the sample Study area WCh Study area WCh

Computer system design, 541511~541519 5471 1002 286 75 scientific research (Tech.) 541711~541720 Management (Mng.) 551111~551114 4438 81 232 62 Legal Services (Legal) 541110~541199 15,036 22 879 13 Administrative, 561110~561990 15,801 151 816 64 Support Services (Admin.) Educational Services 611110~611710 2939 43 276 27 (Edu.) Health Care and Social 621111~624410 11,758 124 729 89 Assistance (Health) Entertainment 713110~713990 Amusement, Gambling 1008 175 335 172 (Etm.) and Recreational Industries Lodgings 721110~721310 871 22 160 18 (Lodging) Restaurants 722110~722410 11,875 371 939 257 (Restr.) Personal Services 812111~812990 6536 137 472 87 (P.Svc.) Non-profit, Public 813110~928120 Religious, Grant-making, 5063 57 359 22 Services Civic Professionals, Public (Pb.Svc.) Administration Others 35,440 583 1889 282 Total 265,650 6054 17,054 4186 49