Cultural Amenities and Unemployment in Dutch Cities: Disentangling a Consumerist and Productivist Explanation for Less-educated Urbanites’ Varying Unemployment Levels across Urban Economies

Abstract

Previous studies on the advanced economies have shown that unemployment levels of the less-educated are low in cities with high employment shares in the advanced producer services. Scholars have consistently interpreted this finding according to Sassen’s polarisation thesis. This article confronts this production-based interpretation with a competing consumption-based explanation, which focuses on the role of cultural amenities. Analysing data on 22 Dutch metropolitan agglomerations between 1996 and 2008, it is shown that, in line with consumerists notions of Richard Florida, advanced producer services settle in cities that are rich in cultural amenities, instead of the other way around as classical labour-follows-capital explanations imply. Yet, only advanced producer services prove to reduce the unemployment levels of the less-educated, meaning that the polarisation thesis, and not the theory on cultural amenities, is a valid explanation for the low level of unemployment in post-industrial cities.

1. Introduction

Production-based theories argue that urban economic development and vitality above all result from the production directed at non-local markets, and assume that labour market and population changes mainly result from the labour demand that this production base yields. A latter-day version of such a production-based theory is Sassen’s polarisation thesis (2001, 2006), which argues that the hallmark of post-industrial urban development in the advanced economies—the clustering of advanced producer services—primarily serves the co-ordination of global production chains. This clustering is in turn assumed to alter the type of urban labour demand in comparison with that of the industrial era: demand for highly educated professionals and less-educated service workers rises while that for industrial workers falls.

According to this line of reasoning, the increasing salience of cultural amenities and consumer services in cities, as for instance observed by Clark (2003b) and Florida (2005), is merely derivative of the clustering of advanced producer services: the ‘consumption base’ caters to the professionals employed in the ‘production base’. However, from the turn of the century onwards, theorising on the consumption base of cities has increasingly challenged production-based theorising when it comes to post-industrial urban economic development (Clark, 2003a; Florida, 2004; Glaeser et al., 2001). Florida (2004), for instance, reverses the productivist logic of Sassen’s polarisation thesis, by arguing that advanced producer services settle in cities that are rich in cultural amenities because those are the places where their potential employees prefer to live. And although not all consumption-based perspectives take such a ‘radical’ view, they at least consider cultural amenities an additional driving force for urban economic development next to the production base, instead of arguing that the former is a mere derivative of the latter. Contrary to the polarisation thesis, those perspectives suggest that the presence of cultural amenities instead of the advanced producer services underlies the alteration of labour demand.

In what follows, I will discuss both aforementioned theoretical positions and their implications for labour demand in more detail. More specifically, the implications of both positions for the demand for less-educated labour will be systematically compared. Subsequently, I will investigate the empirical validity of those implications for the Dutch case, as recent studies interpreted the variation in unemployment levels of the less educated among Dutch cities in accordance to Sassen’s polarisation thesis (van der Waal, 2010b, 2011b; van der Waal and Burgers, 2009, 2011). Consumption-based theorising, however, cast doubts on that interpretation. Unravelling the explanatory value of both perspectives calls for an empirical assessment that is informed by both theoretical positions.

2. The Polarisation Thesis: Advanced Producer Services and Demand for Less-educated Labour

Both Sassen’s polarisation thesis (2001, 2006) and Hamnett’s professionalisation thesis (1994, 1996, 2004) largely focus on the causes and consequences of the clustering of advanced producer services. Their basic idea is that those services—such as finance, law firms, accountancy and consultancy—cluster in cities to produce the inputs that the headquarters of firms need for managing their regional, national or global production chains. According to those two theses, the clustering of such services has been an important driving force for urban economic growth from the 1980s onwards and yielded high demand for professionals—i.e. high-skilled service workers such as financial specialists, accountants and consultants.

Notwithstanding their mutual and undisputed expectation that the clustering of advanced producer services spawns high demand for highly educated professionals, the polarisation thesis and professionalisation thesis radically differ in their expectations on demand for less-skilled labour. While according to the latter the clustering of advanced producer services will hardly generate such demand (Hamnett, 1994, 1996, 2004), the former argues that it creates demand for “an army of low-wage workers” (Sassen-Koob, 1985, p. 262; see also: Sassen, 1991, 2001, 2006), that caters to these professionals in two analytically distinct ways. In the first place ‘directly’, as many low-skilled urbanites find employment in the advanced producer services, or in services that cater to those firms, such as cleaners, clerks and security officers (Sassen, 2000, p. 142; Sassen, 2006, p. 197). Secondly, there is mention of an ‘indirect effect’, sometimes referred to as a ‘multiplier effect’ (Burgers and Musterd, 2002, p. 409), as the consumption patterns and lifestyles of the professionals employed in the advanced producer services are assumed to result in high demand for low-skilled workers in the personal services and the hotel and catering industry.

Although the polarisation thesis was initially formulated for a limited number of global cities, it is important to note that its central claims are also considered valid for cities lower in the urban hierarchy, because

parallel developments exist in cities that function as regional nodes—that is, at smaller geographical scales and lower levels of complexity than global cities (Sassen, 2006, p. 193; see also Sassen, 2000, p. 139 and 2006, p. x).

As a result, it has become standard research practice to assess the impact of the clustering of advanced producer services on labour demand (for an overview, see: van der Waal, 2010b) and income inequality (Zhong et al., 2007) in cities in the advanced economies in general by means of those claims. Two decades of empirical studies on that impact resulted in findings that can be interpreted according to the theoretical rationale of the polarisation thesis, and consequently not according to Hamnett’s professionalisation thesis.

Cities with a strong and growing presence of advanced producer services indeed create ample jobs for the highly as well as the lowly educated, while cities where employment in those services lags behind have to contend with high unemployment levels at the bottom of the labour market. This has been documented for cities in the US (Elliott, 1999, 2004; Kasarda, 1985; Kasarda and Friedrichs, 1985), for cities in the former Federal Republic of Germany (Kasarda, 1985; Kasarda and Friedrichs, 1985) and, most recently, for cities in the Netherlands (van der Waal, 2010b, 2011b; van der Waal and Burgers, 2009, 2011). That this pattern can be found in different institutional systems on both sides of the Atlantic furthermore suggests that it is not the result of national peculiarities, but that it represents a more general process occurring in all advanced economies. With these findings in mind it comes as no surprise that many (local) governments are now considering facilitating the clustering of advanced producer services as the key solution in their battle against unemployment (van der Waal, 2010b).

3. A Consumerist Alternative: Cultural Amenities and Demand for Less-educated Labour

Although the theoretical logic outlined in the previous section gained in prominence from the early 1990s onwards when it comes to understanding labour market opportunities for less-educated urbanites, an alternative logic can be deduced from the more recent literature that focuses on the consumption base, instead of the production base, of post-industrial cities. The scholarly attention paid to the causes and consequences of the local production and consumption of culture and the arts has increased considerably in recent years (see Aoyama, 2009; Clark, 2003a; Currid, 2010). And notwithstanding the broad range of topics covered by those studies, their common denominator is the assumption that the growing salience of cultural amenities in many cities underlies economic development. That denominator challenges production-based notions on urban economic development, which focus on the production of goods and services for non-local markets. As Aoyama puts it

geographical variations across cities have long been explained through processes of production, and seldom through processes of consumption, [while] consumption has become the most prominent feature of postindustrial urban space (Aoyama, 2009, p. 341; see also Markusen and Schrock, 2009, p. 344).

Whether consumption is indeed the most prominent feature of post-industrial cities nowadays remains an empirical question, but the idea that it plays a vital role in urban development has an increasing number of proponents. Not that long ago, Clark also prompted other scholars in urban studies to take consumption and entertainment seriously, as “old paradigms—such as ‘land, labor, capital, and management generate economic development’—are too simple” (Clark, 2003a, p. 1). And recently Markusen and Schrock emphasised, after critically scrutinising studies that focus on production-base explanations

that the consumption base of urban regions—i.e. goods and services that are locally produced and locally consumed, can also be a source of regional job growth and stability (Markusen and Schrock, 2009, p. 344).

These ideas provide a stepping stone to a consumption-based alternative for the production-based explanation for varying unemployment levels across post-industrial cities provided by the polarisation thesis outlined earlier.

They do so because the cultural industries, just like the advanced producer services, might also yield high demand for less-educated labour by means of two mechanisms. The first one is that those industries are likely to employ a substantial number of low-skilled workers to facilitate its main processes. Just like office buildings, many cultural activities require cleaning and security and, of course, the visitors that those activities attract also need to be catered to (see Clark, 2003b; Florida, 2004). The second mechanism that relates cultural amenities to demand for less-educated labour is the consumption pattern of the professionals employed in the cultural industries. As it is suggested that their “propensity to spend locally is high, because of the structure of their preferences” (Markusen and Schrock, 2009, p. 345). More specifically, it is assumed that their consumption pattern roughly mirrors that of the professionals in the advanced producer services, yielding high demand for personal services and the hotel and catering industry. Although both mechanisms provide an additional explanation for the varying unemployment levels of the lowly educated among cities to the two provided by the polarisation thesis—they can exist side-by-side—the former might (partly) account for the variation in those unemployment levels that has previously been attributed to the latter.

In the first place, cities that are renowned for what they have to offer in culture and the arts, almost without exception also stand out as hubs for advanced producer services—take, for instance, New York, London, Paris and Amsterdam. And studies on the US have shown that these empirical observations are more than anecdotal evidence: cities with high employment shares in the advanced producer services are also the ones that are rich in cultural amenities and this does not merely apply to the already-mentioned usual suspects at the top of the urban hierarchy (Currid and Connolly, 2008; Florida, 2003, 2005). This suggests that the employment effects of the clustering of advanced producer services at the bottom of urban labour markets might be overstated: the demand for low-skilled workers, which has thus far been attributed to that clustering by means of the polarisation thesis, might (partly) stem from the local cultural consumption base as implied by the theory on cultural amenities developed in this section.

Secondly, Richard Florida even goes one step further than emphasising that cultural consumption can in itself be of importance for urban economic development, by claiming that the causal direction between labour and capital flows has been radically reversed in the post-industrial era (Florida, 2002, 2003, 2004, 2005). Whereas the productivist claim is that people settle in places with ample job opportunities, he states that many service-production firms settle where young knowledge workers prefer to live. This idea adds up to the consumerist explanation for varying unemployment levels across cities, as Florida furthermore argues that those knowledge workers prefer to live in cities that are rich in cultural amenities (see also Glaeser et al., 2001; Glaeser et al., 2003; Currid, 2010). And this is not a mere theoretical expectation: the share of college graduates in the population is high in cities with an abundance of “constructed amenities, like the numbers of opera, research libraries, used and rare books stores, juice bars, Starbucks, and bicycle events” (Clark, 2003b, p. 132).

The previously mentioned substantial positive relationship between the advanced producer services and cultural amenities, and Florida’s claim that those services settle in places that are rich in those amenities, both indicate that previous interpretations of varying demand for less-educated labour across cities in the advanced economies by the polarisation thesis (see: Elliott, 1999, 2004; Kasarda, 1985; Kasarda and Friedrichs, 1985; van der Waal, 2010b, 2011b; van der Waal and Burgers, 2009, 2011) might be overstated. More specifically, the low unemployment level among less-educated urbanites in cities with a high share of advanced producer services might (partly) be the result of the consumption base of those cities—i.e. the cultural industries—instead of, as previously assumed, their production base, for two reasons. In the first place, because previous interpretations by means of the polarisation thesis (partly) rest on a spurious relationship due to the strong positive relationship between cultural amenities and advanced producer services. (Part of) the relationship between unemployment levels and employment levels in the advanced producer services might, hence, be the result of the presence of cultural amenities as theorised in this section. Secondly, because (part of) the effect of the advanced producer services is in the end driven by the presence of cultural amenities, for cultural amenities underlie the settlement of those services in the first place as argued by Florida. Hence, for an assessment on the empirical validity of the productivist polarisation thesis and the alternative consumerist one developed in this section for explaining varying unemployment levels across cities, both need to be modelled simultaneously, as will be done in what follows. Such an assessment, however, calls for modelling a third explanation for the strong presence of advanced producer services and cultural amenities next to the already-mentioned ones which claim that one underlies the other: the existence of agglomeration economies.

Contrary to both the classical labour-follows-capital explanation and Florida’s capital-follows-labour one, advanced producer services and cultural amenities might be related because both, for various and different reasons, can reap economic benefits from the mere size of a city, such as knowledge spillovers, a large supply of consumers, good infrastructure (Fujita and Thisse, 2002) and information and communication loops (Storper and Venables, 2004). Although the exact mechanisms by means of which advanced producer services and cultural industries will benefit from settlement within the largest cities are not the focus of this article, disentangling the already-mentioned causal relationship between the presence of those services and industries at least calls for controlling for agglomeration economies. It might be that the previously demonstrated relationship between the two is not because one results from the other, but that both are the result of agglomeration economies, and can hence especially be found in the largest cities. This would explain a strong relationship between advanced producer services and cultural amenities even if these two are not causally related as claimed in the production- and consumption-based theories.

Figure 1 illustrates the conceptual model that integrates the theoretical arguments already outlined. The left-hand side depicts the expectations on the basis of agglomeration effects. The vertical vector on the left side indicates the expectation deduced from Florida’s consumerist notion that the growth of the advanced producer services is caused by the presence of the cultural industries (hypothesis 1). The vertical vector on its right-hand side, on the other hand, represents the expectation deduced from the classical labour-follows-capital or productivist notion that cultural industries develop in places with high shares of advanced producer services (hypothesis 2).

Figure 1.

The expected impact of 1) the working population on the advanced producer services, cultural industries and hotel and catering industry (agglomeration effect, 2) the advanced producer services on the cultural industries and/or vice versa (production-driven and/or consumption-driven settlement), and 3) the advanced producer services (polarisation thesis, or production-based explanation) and cultural industries (thesis on cultural amenities, or consumption-based explanation) on the employment level of less-educated urbanites.

The upper right-hand side of Figure 1 represents the polarisation thesis. It predicts that the advanced producer services have a direct negative effect on the unemployment level of less-educated urbanites (hypothesis 3), due to the high demand they yield for cleaners, security workers, servants and the like. It is furthermore expected that those services have an indirect negative effect on those unemployment levels because the consumption patterns and lifestyles of the urban professionals they employ yield high demand for low-skilled workers in the personal services and hotel and catering industry (hypothesis 4).

The lower right-hand side depicts the consumption-based explanation developed in this section, which predicts a direct negative effect of the cultural industries on the unemployment level of less-educated urbanites (hypothesis 5), due to high demand for less-educated service workers such as cleaners, security workers and servants. Also, it is expected that the cultural industries have an indirect negative effect on those unemployment levels because the consumption patterns and lifestyles of the urban professionals they employ yield high demand for low-skilled workers in the personal services and hotel and catering industry (hypothesis 6).

The hypotheses will be tested by comparing Dutch metropolitan areas (which in what follows will be referred to as ‘cities’), as the varying unemployment levels among those have recently been interpreted according to the polarisation thesis (van der Waal, 2010b, 2011b; van der Waal and Burgers, 2009, 2011). The Netherlands provides a strategic case for that comparison, as its welfare policies are highly centralistic in comparison with those of other Western countries, especially since the 1980s when most European countries decentralised many (financial) responsibilities to local governments. As a result, the increased differentiation between cities when it comes to welfare measures and labour market interference characteristic of other European countries in recent decades, is largely absent in the Netherlands (Burgers and Musterd, 2001; Musterd et al., 1998; Newman and Thornley, 1996; Parkinson et al., 1988). Hence, different unemployment levels among Dutch cities do not result from different types of labour market regulation. In short, the analyses that follow keep labour market regulation constant by design.

4. Data and Operationalisation

To test the central expectations of this study, I have constructed a dataset on the 22 metropolitan areas in the Netherlands, using data available through the website of ‘Statistics Netherlands’ (CBS).¹ Although these data are available from the early 1990s until 2008, for some variables addressed here the data did not fully cover this time-span and for some analyses there is a need to model time-lags. As a result, the analysis of the causal relationship between the advanced producer services and the cultural industries includes 22 Dutch metropolitan areas between 1997/98 and 2007 (242/220 city/year combinations), while the analyses on their impact on unemployment levels include these areas between 1998 and 2007 (220 city/year combinations).

Unemployment less educated measures the share of the less-educated working population that is currently unemployed, but is looking for a job for at least 12 hours a week—the standard Dutch definition of unemployment. It is measured two years later than the independent variables, because the unemployment effects of structural changes in the economy will not show themselves immediately.

Advanced producer services measures the share of the working population that is employed in firms classified in class J (finance) and class K (real estate and producer services) in the Dutch SBI 93 classification (Standaard Bedrijfsindeling 1993) that corresponds to the ‘ISIC Rev. 3.1’ (International Standard Industrial Classification of All Economic Activities) of the United Nations. Put differently, advanced producer services indicates the presence of the activities that, according to the polarisation thesis, are the primary drivers for job growth at the bottom of the labour market. In the first place, this is because these services are thought to yield high demand for lower-educated workers themselves. In the second place, this is because it is argued that the presence of these services indirectly leads to high demand for these workers in other industries (see later).

Cultural industries measures the share of the working population that is employed in firms classified in class O (recreational, cultural and sporting activities) in the Dutch SBI 93 classification (corresponding to the ‘ISIC Rev. 3.1’). Hence, it indicates the extent to which an urban economy functions as an entertainment machine. As a result, it provides a good indicator for testing the consumption-based explanation for varying unemployment levels developed in the previous section.

On the basis of both the polarisation thesis and the theory on cultural amenities, it can be expected that part of the high labour demand for low-skilled urbanites induced by the advanced producer services and cultural industries respectively, is accounted for by the hotel and catering industry. This variable measures the share of the working population that is employed in firms classified in class H (hotels and restaurants; ‘ISIC Rev. 3.1’). In line with both Sassen’s polarisation theory and its consumption-based alternative, it would be optimal if, in addition to modelling the impact of the hotel and catering industry, the impact of the share of employment in private households (class T in the ‘ISIC Rev. 3.1’) could also be modelled. Unfortunately, the latter is not available at the metropolitan level.

Manufacturing indicates the share of the working population that is employed in firms classified in class D (manufacturing) in the Dutch SBI 93 classification (corresponding to the ‘ISIC Rev. 3.1’). It is used to control for Wilson’s mismatch thesis (1996) which, aptly put, claims that urban unemployment results from deindustrialisation and that cities with an industrial legacy—i.e. with high scores on this variable—will consequently have to deal with high unemployment levels among their less-educated population.

There is a need for several additional control variables since the population in metropolitan areas is not completely determined by the labour market, and agglomeration effects might underlie the presence of advanced producer services and cultural industries, as well as unemployment levels.

The four variables Age 15–24/25–34/35–44/45–54 measure the share of the working population by age category. These are used to control for non-labour-market-driven settlement patterns, such as the high share of students in university cities, that might affect the unemployment levels of less-educated urbanites.

Immigrant share measures the share of immigrants in the population of a metropolitan agglomeration. This needs to be controlled for as previous studies have shown that, in accordance with the substitution thesis, Dutch cities with high immigrant shares have to contend with high unemployment levels among their less-educated citizens due to labour-market substitution (van der Waal, 2011b).

Share less educated measures the share of the working population that is less educated. It is used to control for non-market-driven settlement patterns. The share of cheap subsidised housing for instance differs by city, but is likely to be highly related to the share of less-educated urbanites and hence to their unemployment levels.

Working population measures the total number of urbanites between the age of 15 and 66, either employed or unemployed, indicating the total working population. This variable is entered in logarithmic form due to its skewed distribution. In the first place, it is used as a control for agglomeration effects in the analysis of the causal relationship between the advanced producer services and the cultural industries (see Meijers, 2008). In the analysis of the unemployment levels of less-educated urbanites, it is in the second place used to control for the effect of city size on unemployment levels, attributed to urbanisation economies (see Alperovich, 1993), and the concomitant industrial diversity and risk diversification (Gan and Zhang, 2006).

5. Results

Before testing the hypotheses formulated earlier, I will first provide insights into the relationship between the working population, advanced producer services, the cultural industries and the hotel and catering industry. These are shown in Table 1 by means of a partial correlation—controlled for measurement year—to account for the hierarchical structure (years within cities) in the data. It shows that, just as in the US (Currid and Connolly, 2008; Florida, 2003, 2005), cities in the Netherlands with a high share of employment in the advanced producer services also have high employment shares in cultural industries. Furthermore, and in accordance with the polarisation thesis, high levels of employment in the advanced producer services go hand-in-hand with high employment shares in the hotel and catering industry. This suggests that the professionals employed in the advanced producer services indeed yield high demand for the hotel and catering services.

Table 1.

Partial correlation (controlled for year) between working population, advanced producer services, cultural industries and hotel and catering industry in 22 Dutch metropolitan agglomerations, 1996–2007 (N = 264)

	(1)	(2)	(3)	(4)
Working population (1)	1.000	0.482***	0.205***	0.390***
Advanced producer services (2)		1.000	0.403***	0.225***
Cultural industries (3)			1.000	0.023
Hotel and catering industry (4)				1.000

Notes: *p <0.10; ** p <0.05; *** p <0.01. Analyses of own dataset computed with data retrieved from Statline service, Statistics Netherlands (CBS).

Contrary to what could be expected on the basis of the consumption-based counterpart of the polarisation thesis, which focuses on cultural amenities, an association between the hotel and catering industry and the cultural industries does not exist. Table 1 furthermore suggests that both the advanced producer services and the cultural industries reap the benefits of agglomeration economies: the strongest presence of those services and industries can be found in the largest cities. Assessing the causal relationships between the advanced producer services, cultural industries and the working population, however, calls for another analysis that is provided in Table 2.

Table 2.

Multilevel regression analysis: dependent variable is the share of producer services and cultural industries in the 22 Dutch metropolitan agglomerations, 1997–2007 (method: maximum likelihood, entries are regression coefficients)

	Producer services			Cultural industries
	Null model	Model 1 (t – 1)	Model 2 (t – 2)	Null model	Model 1 (t – 1)	Model 2 (t – 2)
Independents	β	β	β	β	β	β
Constant	0.000	0.000	−0.000	0.000	0.000	0.000
Working population		0.788***	0.607***		0.312*	0.399**
Cultural industries		0.058	0.157**
Producer services					−0.020	0.014
Variance year level	0.072***	0.062***	0.066***	0.086***	0.084***	0.076***
Variance metropolitan level	0.923***	0.896***	0.761***	0.910***	0.875***	0.903***
Deviance	153.97	123.98	126.00	186.82	182.74	153.79
N	242	242	220	242	242	220

Notes: *p <0.10; ** p <0.05; *** p <0.01. Analyses of own dataset computed with data retrieved from Statline service, Statistics Netherlands (CBS).

Although uncovering a metaphysical construct like causality is strictly speaking not possible in a statistical analysis, a necessary condition that needs to be met for such causality to exist is that causes precede effects. The most obvious way to bring such chains of events to light by means of a statistical analysis is to model time-lags, as has recently been done by Neal (2012), when tackling a similar causality problem. Besides modelling such time-lags, there is furthermore need to control for autocorrelation as the data used are hierarchical: years nested within cities. If such autocorrelation is not controlled for, the number of cases (cities^*years) is artificially increased, which leads to an underestimation of standard errors and consequently yields artificially high significance levels. Therefore multilevel modelling will be applied, which estimates the variance in the dependent variable at year and city level separately. Although a fixed-effects regression model can also control for autocorrelation by entering city and year dummies, multilevel modelling has the additional advantage that it is not “extremely conservative” by “throwing out” all between-city and between-year variation in the data (Alderson and Nielsen, 2002, p. 1269). Also, the latter method is able to model variables that are (nearly) constant—for instance: a (nearly) equal share of employment in a given industry across the assessed time-span—while this would in the former method lead to multicollinearity problems as that variable would be (nearly) equal to the city-level dummy that it needs to include, as to control for autocorrelation.

The first three columns in Table 2 test hypothesis 1: advanced producer services settle in cities that are rich in cultural amenities. The first column shows that the variance in the producer services largely exists at city level (0.923/(0.923 + 0.072) = 92.8 per cent): the bulk of the variance in employment in the advanced producer services can hence be attributed to city-level differences, and not by differences in time. Models 1 and 2 test hypothesis 1 by modelling a time-lag of one and two years respectively—the scores on the independent variables in the models are one and two years prior to those in the dependent variable—while controlling for agglomeration effects.² Both show that the presence of advanced producer services strongly depends on these agglomeration effects: the coefficient of working population is positive and significant. However, there is also some indication that those services prefer to settle in cities that are rich in cultural amenities. There is a positive relationship between the employment in the cultural industries and that in the producer services one and two years later, albeit that only in the latter case it is significant. This corroborates hypothesis 1.

Columns 4–6 test hypothesis 2: employment in the cultural industries depends on that in the advanced producer services. The null model shows that the bulk of the variance in the employment in the cultural industries, just as the variance in the employment in the advanced producer services, exists at the city level (0.910/(0.910 + 0.086) = 91.3 per cent). The former furthermore partly depends on agglomeration effects, albeit less so than the latter. There is, however, no empirical evidence that supports hypothesis 2: employment in the cultural industries is not related to employment in the advanced producer services one or two years earlier.³ In short, the analyses in Table 1 point in the direction of Florida’s argument that advanced producer services settle in places that are rich in cultural amenities. Note, however, that according to the analysis in Table 2 the extent to which the presence of the latter determines that of the former is rather small. The classical notion that cultural consumption is in the end derivative of a city’s production base, on the other hand, does not seem to apply to Dutch cities at all.

The question is how these findings relate to interpreting low unemployment levels in cities with high employment shares in the advanced producer services by means of the polarisation thesis. This will be assessed in Table 3 by means of a multilevel analysis of the unemployment levels of the less-educated across cities and years. Next to the advantages for dealing with hierarchical data already outlined, multilevel modelling is of additional value for assessing unemployment levels as these also depend on economic booms and busts. That variation will, hence, largely occur at year level: during economic busts (booms) the unemployment level of all cities will be higher (lower), while city-level differences remain approximately equal (van der Waal, 2011a). The variation in unemployment levels among cities due to their production and consumption base are of interest here and will, on the other hand, largely occur at city level (descriptives on unemployment levels can be found in the appendix).

Table 3.

Multilevel regression analysis: dependent variable is unemployment less educated in 22 Dutch metropolitan agglomerations, 1998–2007 (method: maximum likelihood, entries are regression coefficients; N = 220 (22 cities * 10 years))

Independents	Null model	Model 1	Model 2	Model 3
	β	β	β	β
Constant	0.000	0.000	−0.000	−0.000
Producer services		−0.273**	−0.304**	−0.309**
Cultural industries			0.081	0.087
Hotel and catering industry				0.036
Controls
Manufacturing		−0.046	−0.032	−0.024
Working population		−0.167	−0.154	−0.161
Share lower educated		−0.247**	−0.252**	−0.249**
Immigrant share		0.532***	0.512***	0.510***
Age 15–24		0.230	0.217	0.223
Age 25–34		−0.038	−0.043	−0.033
Age 35–44		0.114	0.116	0.120
Age 45–54		−0.002	−0.008	−0.008
Variance year level	0.779***	0.606***	0.599***	0.596***
Variance metropolitan level	0.217**	0.202	0.232	0.238
Deviance	598.60	546.52	546.01	546.00

Notes: *p <0.10; ** p <0.05; *** p <0.01. Analyses of own dataset computed with data retrieved from Statline service, Statistics Netherlands (CBS).

The null model of Table 3 first shows that about four-fifths of the variance in the unemployment levels among less-educated urbanites are accounted for by differences in time (0.779/(0.779 + 0.217) = 78.2 per cent) and one-fifth by differences between cities (0.217/(0.779 + 0.217) = 21.8 per cent). Model 1 yields results similar to those of previous studies: unemployment levels are highest in cities with high immigrant shares (van der Waal, 2010a) and, as could be expected on the basis of the polarisation thesis, lowest in cities with high employment shares in the advanced producer services (van der Waal, 2010b; van der Waal and Burgers, 2009, 2011).⁴

Does the latter association indeed need to be interpreted according to the polarisation theory? This remains an empirical question for it might (partly) be the result of the high employment share in the cultural industries, which would mean that the theory on cultural amenities can (partly) account for it. Model 2, however, clearly indicates that the latter scenario does not hold for contemporary Dutch cities. The employment share in the cultural industries does not affect the unemployment levels among less-educated urbanites whatsoever, as its insignificant coefficient shows.⁵ And even if it were significant, it would be contrary to what one would expect on the basis of the theory on cultural amenities: it is positive instead of negative. Hence, hypothesis 3, deduced from the polarisation thesis, is corroborated, while hypotheses 5 and 6, deduced from the theory on cultural amenities, need to be rejected.

The last model is dedicated to finding out whether—and, if so, to what extent—the employment in the hotel and catering industry can account for the low unemployment levels in cities with high employment shares in the advanced producer services.⁶ Contrary to what could be expected on the basis of the polarisation thesis, hotel and catering industry does not yield a significant coefficient and consequently cannot account for (part) of the low unemployment levels in cities where the advanced producer services cluster most strongly—rejecting hypothesis 4. All in all, the findings indicate that the polarisation thesis partially—that is, as far it concerns the direct labour demand for low-skilled service workers due to the clustering of advanced producer services (upper path in Figure 1)—has explanatory value for the varying unemployment levels among less-educated Dutch urbanites. Its alternative on cultural amenities, on the other hand, is not empirically corroborated. The concluding section will further elaborate on these findings.

6. Conclusions and Debate

In recent years, many studies have been devoted to emphasising the crucial role of consumption in latter-day urban economies and to formulating theses as to how this role can be understood. The reason to do so, is that

studies of the city traditionally posit a division between a city’s economy and its culture, with culture subordinate in explanatory power to ‘work’ (Clark et al., 2003, p. 291).

For the problem at hand, varying job opportunities for less-educated urbanites in contemporary Dutch cities, a production-based explanation, however, still seems to be in place. The only expectation deduced from the consumption-based theorising that was corroborated in this study is that advanced producer services firms settle in places that are rich in cultural amenities. Although this is in line with Florida’s most radical claims, it needs to be interpreted with care because the effect is very small. A possible explanation for this small effect is that the clustering of cultural industries and/or advanced producer services is, besides the result of agglomeration effects, mainly driven by a clustering (Porter, 1998) and/or network (Grabher and Powell, 2004) logic. This is, however, for future research to decide and less relevant for the central focus of this article: the explanation for the varying unemployment levels of less-educated urbanites across Dutch cities.

These could be explained by the polarisation thesis, and not by the competing theory on cultural amenities—albeit that there merely proves to be a direct effect of the advanced producer services on those unemployment levels, and not an indirect one via the consumption pattern and lifestyles of professionals as measured by the hotel and catering industries. This suggests that contemporary Dutch urban labour markets are still primarily production-driven. Only in as far as the settlement of advanced producer services is driven by the presence of cultural amenities—which in the Dutch case proves to be marginally so— could one argue that the consumption base of Dutch cities is also indirectly responsible for varying unemployment levels of less-educated urbanites across cities.

All that said, the findings do not imply that the consumption-based explanations for current economic developments in cities in the advanced economies need to be discarded altogether. On the contrary, this study merely sheds light on the relationship between one production-based explanation for explaining varying unemployment levels across cities and a consumption-based alternative. And while a focus on the Netherlands is useful as it keeps labour market regulation constant by design, three particularities of this study suggest that much work still needs to be done as to uncover the complexities of, first, that specific relationship in the advanced economies in general and, secondly, the relationship between production-based explanations and their consumption-based alternatives for economic developments less narrowly defined, such as economic and employment growth.

In the first place, the Netherlands has a relatively generous welfare regime and a relatively strongly regulated labour market that accompanies it. As a result, it is a context where the consumption patterns of the higher-educated are least likely to boost demand for low-skilled service workers, because the less-educated are less dependent on the labour market for providing themselves with the means of subsistence than those in many other developed economies. Furthermore, the relatively high minimum wage levels in the Netherlands are likely to hamper the creation of many service activities in the cultural industries, hotel and catering industry, and in personal services. That the polarisation thesis is corroborated only in so far as it concerns the direct impact of the advanced producer services on demand for less-skilled labour, and not when it comes to the consumption pattern of the professionals those services employ, might, hence, be a Dutch particularity that can be attributed to its welfare institutions. And the same goes for the finding that the growth of the cultural industries is not driven by the presence of the advanced producer services, as part of the cultural industries in the Netherlands is subsidised, and their place of settlement might, hence, not be one-on-one related with the presence of their consumers—highly educated professionals—as suggested by the classical production-based logic addressed in this article.

Secondly, the Netherlands is a particularly small country. The production and consumption of cultural amenities in cities other than the ones in which urbanites live and/or work, are therefore options that are largely absent in larger developed economies. As a result, the potential job openings that cultural production and consumption provide for less-educated workers might be less place-bound in the Netherlands than consumption-based theories of urban development account for. Hence, local unemployment levels and local employment in cultural industries do not necessarily need to be empirically related in the Netherlands. The two aforementioned Dutch peculiarities imply that in bigger and less-regulated advanced economies—most notably the US—the presence of cultural amenities can go hand-in-hand with high demand for low-skilled service workers as consumption-based explanations imply.

Thirdly, cultural amenities have been measured with a rather crude measure for the issue at hand, as the indicator used also entails the production of cultural goods directed at the non-local market. There is, however, unfortunately no less crude measure of cultural production available at the city level. All in all, although a comparison of Dutch cities is, on the one hand, ideal for a comparison of unemployment levels as it controls for local variation in labour market intervention, the findings of this study do, on the other hand, not necessarily travel beyond the Dutch case because of its generous welfare regime, its public funding of the cultural industries and the crude measure for the cultural amenities at the city level.

In addition, the rejection of the consumption-based explanation needs to be related to the narrow focus of this article: varying unemployment levels of the less-educated across Dutch cities. This in no way disqualifies consumption-based explanations that focus on the impact of cultural amenities on other economic developments. The bulk of the literature on such explanations on the one hand revolves around more encompassing phenomena such as population growth as measured in the increase in the urban population in general (Glaeser et al., 2001; Glaeser and Gottlieb, 2006), or of more narrowly defined groups such as the creative class (Florida et al., 2008) or bohemians (Wojan et al., 2007). On the other hand, consumption-based explanations focus on the impact of cultural amenities on general measures of economic vitality and development (see Currid, 2009; Florida et al., 2008; Wojan et al., 2007). Many of these studies, mainly focusing on the US, have empirically validated such consumption-based explanations, and those findings do not need to be at odds with those of the present study.

Finally, I would like to make some remarks on urban policies aimed at battling unemployment inspired by the polarisation thesis and its alternative that revolves around cultural amenities. As has been previously mentioned by Peck (2005), the enthusiasm of policy-makers for the production and subsequent utilisation of urban culture for employment strategies might be premature and unwarranted. And the same goes for the idea that the settlement of advanced producer services can be manipulated and subsequently be utilised to that same end (see van der Waal, 2010a). First, and completely in line with this study’s findings, empirical evidence for the idea that “cappuccino urban politics” (Peck, 2005, p. 760) can remedy unemployment is scant. Secondly, the settlement of advanced producer services is difficult to manipulate: while low-priced office space is widely available in many Dutch cities, numerous advanced producer services firms still opt for a limited number of very expensive places, with Amsterdam’s Zuidas as the primary example. Thirdly, as far as urban policies inspired by the theoretical notions that guided this research succeed when it comes to battling unemployment, they are likely to intensify class divisions and underemployment (see Peck, 2005). For all those involved in policy decisions, awareness of these effects can be informative and valuable for their future considerations.

Footnotes

Appendix

Table A1.

Averages and standard deviations of unemployment less educated in 22 Dutch metropolitan agglomerations, 1998–2007 (N = 22)

	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007
Average	4.45	3.65	3.01	2.83	3.35	4.66	5.68	5.82	5.07	4.33
Standard deviation	1.50	1.38	1.27	1.07	1.03	1.38	1.40	1.87	1.54	1.35

Source: own dataset computed with data retrieved from Statline service, Statistics Netherlands (CBS).

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Alderson

A. S.

Nielsen

(2002) Globalization and the great U-turn: income inequality trends in 16 OECD countries, American Journal of Sociology, 107(5), pp. 1244–1299.

Alperovich

(1993) City size and the rate and duration of unemployment: evidence from Israeli data, Journal of Urban Economics, 34(3), pp. 347–357.

Aoyama

(2009) Introduction: consumption-centered research for diverse urban economies, Urban Geography, 30(4), pp. 341–343.

Burgers

Musterd

(2001) Global and local determinants of social exclusion: Amsterdam versus Rotterdam, in: Andersen

H. T.

van Kempen

(Eds) Governing European Cities: Social Fragmentation, Social Exclusion and Urban Governance, pp. 125–149. Aldershot: Ashgate.

Burgers

Musterd

(2002) Understanding urban inequality: a model based on existing theories and an empirical illustration, International Journal of Urban and Regional Research, 26(2), pp. 403–413.

Cheshire

(1990) Explaining the recent performance of the European Community’s major urban regions, Urban Studies, 27(3), pp. 311–333.

Clark

T. N.

(2003a) Introduction: taking entertainment seriously, Research in Urban Policy, 9, pp. 1–17.

Clark

T. N.

(2003b) Urban amenities: lakes, opera, and juice bars: do they drive development?, Research in Urban Policy, 9, pp. 103–140.

Clark

T. N.

Lloyd

Wong

K. K.

Jain

(2003) Amenities drive urban growth: a new paradigm and policy linkages, Research in Urban Policy, 9, pp. 291–322.

10.

Currid

(2009) Bohemia as subculture; ‘bohemia’ as industry: art, culture, and economic development, Journal of Planning Literature, 23(4), pp. 368–382.

11.

Currid

(2010) Symposium introduction: art and economic development: new directions for the growth of cities and regions, Journal of Planning Education and Research, 29(3), pp. 257–261.

12.

Currid

Connolly

(2008) Patterns of knowledge: the geography of advanced services and the case of art and culture, Annals of the Association of American Geographers, 98(2), pp. 414–434.

13.

Elliott

J. R.

(1999) Urban restructuring revisited: organizational and labor market changes during the 1980s, Sociological Perspectives, 42(3), pp. 421–438.

14.

Elliott

J. R.

(2004) The work of cities: underemployment and urban change in late-20th-century America, Cityscape, 7(1), pp. 107–133.

15.

Florida

(2002) Bohemia and economic geography, Journal of Economic Geography, 2(1), pp. 55–71.

16.

Florida

(2003) Cities and the creative class, City & Community, 2(1), pp. 3–19.

17.

Florida

(2004) The Rise of the Creative Class: And How It’s Transforming Work, Leisure, Community and Everyday Life. New York: Basic Books.

18.

Florida

(2005) Cities and the Creative Class. New York: Routledge.

19.

Florida

Mellander

Stolarick

(2008) Inside the black box of regional development: human capital, the creative class and tolerance, Journal of Economic Geography, 8(4), pp. 615–649.

20.

Fujita

Thisse

J.-F.

(2002) Economics of Agglomeration: Cities, Industrial Location, and Regional Growth. Cambridge: Cambridge University Press.

21.

Gan

Zhang

(2006) The thick market effect on local unemployment rate fluctuations, Journal of Econometrics, 133(1), pp. 127–152.

22.

Glaeser

E. L.

Gottlieb

J. D.

(2006) Urban resurgence and the consumer city, Urban Studies, 43(8), pp. 1275–1299.

23.

Glaeser

E. L.

Kolko

Saiz

(2001) Consumer city, Journal of Economic Geography, 1, pp. 27–50.

24.

Glaeser

E. L.

Kolko

Saiz

(2003) Consumers and cities, Research in Urban Policy, 9, pp. 177–183.

25.

Grabher

Powell

W. W.

(Eds) (2004) Networks. Cheltenham: Edward Elgar.

26.

Hamnett

(1994) Social polarisation in global cities: theory and evidence, Urban Studies, 31(3), pp. 401–425.

27.

Hamnett

(1996) Social polarisation, economic restructuring and welfare state regimes, Urban Studies, 33(8), pp. 1407–1430.

28.

Hamnett

(2004) Unequal City: London in the Global Arena. London: Routledge.

29.

Kasarda

J. D.

(1985) Urban change and minority opportunities, in: Peterson

P. E.

(Ed.) The New Urban Reality, pp. 33–67. Washington, DC: Brookings Institution Press.

30.

Kasarda

J. D.

Friedrichs

(1985) Comparative demographic-employment mismatches in U.S. and West German cities. Paper presented at the Annual Meeting of the Population Association of America, Boston, MA, March.

31.

Kasarda

J. D.

Friedrichs

(1986) Economic transformation, minorities and urban demographic–employment mismatch in the U.S. and West Germany, in: Ewers

H. J.

Goddard

J. B.

Matzerath

(Eds) The Future of the Metropolis: Berlin, London, Paris, New York Economic Aspects, pp. 221–249. Berlin: Walter de Gruyter.

32.

Markusen

Schrock

(2009) Consumption-driven urban development, Urban Geography, 30(4), pp. 344–367.

33.

Marlet

Woerkens

C. van

(2007) The Dutch creative class and how it fosters urban employment growth, Urban Studies, 44(13), pp. 2605–2626.

34.

Meijers

(2008) Summing small cities does not make a large city: polycentric urban regions and the provision of cultural, leisure and sports amenities, Urban Studies, 45(11), pp. 2323–2342.

35.

Musterd

Ostendorf

Breebaart

(1998) Multi-ethnic Metropolis: Patterns and Policies. Dordrecht: Kluwer.

36.

Neal

(2012) Creative employment and jet set cities: disentangling causal effects, Urban Studies, 49(12), pp. 2693–2709.

37.

Newman

Thornley

(1996) Urban Planning in Europe: International Competition, National Systems and Planning Projects. London: Routledge.

38.

Parkinson

Foley

Judd

(Eds) (1988) Regenerating the Cities: The UK Crisis and the US Experience. Manchester: Manchester University Press.

39.

Peck

(2005) Struggling with the creative class, International Journal of Urban and Regional Research, 29(4), pp. 740–770.

40.

Porter

M. E.

(1998) Clusters and the new economics of competition, Harvard Business Review, November/December, pp. 77–90.

41.

Sassen-Koob

(1984) The new labour demand in global cities, in: Smith

M. P.

(Ed.) Cities in Transformation, pp. 139–171. Beverly Hills, CA: Sage.

42.

Sassen-Koob

(1985) Capital mobility and labor migration: their expression in core cities, in: Timberlake

(Ed.) Urbanization in the World-Economy, pp. 231–265. Orlando, FL: Academic Press.

43.

Sassen-Koob

(1986) New York City: economic restructuring and immigration, Development and Change, 17(2), pp. 85–119.

44.

Sassen

(1991) The Global City: New York, London, Tokyo. Princeton, NJ: Princeton University Press.

45.

Sassen

(2000) Cities in a World Economy, 2nd edn. London: Pine Forge Press.

46.

Sassen

(2001) The Global City: New York, London, Tokyo, 2nd edn. Princeton, NJ: Princeton University Press.

47.

Sassen

(2006) Cities in a World Economy, 3rd edn. Thousand Oaks, CA: Pine Forge Press.

48.

Sassen

(2007) A Sociology of Globalization. New York: W.W. Norton & Co.

49.

Storper

Venables

A. J.

(2004) Buzz: face-to-face contact and the urban economy, Journal of Economic Geography, 4(4), pp. 351–370.

50.

Waal

J. van der

(2010a) Stedelijke economieën in een tijd van mondialisering: Arbeidsmarktkansen en etnocentrisme van laaggeschoolden in Nederlandse steden. Amsterdam: Amsterdam University Press.

51.

Waal

J. van der

(2010b) Unravelling the global city debate: economic inequality and ethnocentrism in contemporary Dutch cities. PhD thesis, Erasmus University Rotterdam.

52.

Waal

J. van der

(2011a) De diensteneconomie als banenmachine voor laaggeschoolden, in: Hamel

Cueva

Jacobs

Nijhof

(Eds) Economische Verkenningen Rotterdam, pp. 100–102. Rotterdam: Ontwikkelingsbedrijf Rotterdam.

53.

Waal

J. van der

(2011b) Post-industrialisation, immigration, and unemployment: how and why the impact of immigration on unemployment differs between Dutch cities, Urban Studies, 49(8), pp. 1711–1724.

54.

Waal

J. van der

Burgers

(2009) Unravelling the global city debate: economic inequality and ethnocentrism in contemporary Dutch cities, Urban Studies, 46(13), pp. 2715–2729.

55.

Waal

J. van der

Burgers

(2011) Post-industrialisation, job opportunities and ethnocentrism: a comparison of 22 Dutch urban economies, Urban Studies, 48(4), pp. 681–697.

56.

Wilson

W. J.

(1996) When Work Disappears: The World of the New Urban Poor. New York: Alfred A. Knopf.

57.

Wojan

T. R.

Lambert

D. M.

McGranahan

D. A.

(2007) Emoting with their feet: bohemian attraction to creative milieu, Journal of Economic Geography, 7(6), pp. 711–736.

58.

Zhong

Clark

T. N.

Sassen

(2007) Globalization, producer services, and income inequality across US metro areas, International Review of Sociology, 17(3), pp. 385–391.