Chris Hamnett. Handbook of Urban Studies. Editor: Ronan Paddison. SAGE Publications. 2001.
Any city, however small, is in fact divided into two, one the city of the poor, the other of the rich: these are at war with one another, and in either there are many smaller divisions, and you would be altogether beside the mark if you treated them all as a single State.
- Plato, RepublicIV, 422B
The existence of concentrations of rich and poor in the world’s major cities is a long-standing phenomenon. Plato commented on it in his Republic, and the social divisions in the growing industrial cities were a perennial subject of debate in the nineteenth century from Engels’s (1849) work of the condition of the English working classes onwards. Although similar concerns continued to surface in Britain during the 1930s, the baton was effectively transferred to the USA in the late nineteenth century (Ward, 1989), first to New York and Boston, and then, in the 1920s to Chicago with the work of Park and Burgess (1925). Concern about poverty, segregation and the inner city then went quiet to a large extent during the 1940s, 1950s and 1960s as sociological interest shifted to the growing suburbs. It was not until the American urban riots of the mid-1960s, and the realization of the magnitude of the racial and ethnic transformation of many American inner cities in the post-war decades, that interest resurfaced in the question of class and segregation (Beauregard, 1993a, 1993b; Fainstein, 1993). The 1960s and 1970s saw a large volume of research on ethnic segregation, particularly in the USA (Morrill, 1965; Rose, 1970), but also in Britain and elsewhere. These decades also saw a great deal of quantitative research on patterns of residential differentiation by geographers and sociologists. In the 1980s, however, two new research themes emerged, the first revolving around the existence of the ‘underclass’ and its structural and behavioural causes, and the second focusing on what is known as social polarization and the related issue of urban ‘duality’ and dual cities. These questions have been linked to issues of race, ethnicity and segregation, (Castells, 1989; Sassen, 1984, 1986) though they are by no means synonymous.
In this chapter I intend to trace the development of these concerns and issues paying particular attention to recent debates concerning polarization, duality and the underclass. The structure of the chapter is broadly historical, though the discussion of earlier work is extremely attenuated, not least because this material is already well known and well documented. My principal focus is on the work done in the past 10-15 years.
Classc Structure and Residential Differentiation in the Nineteenth Century
The publication of Engels’s (1849) The Condition of the Working Classes in England, 1848 drew attention to the growth of the urban industrial working class in Britain, and to the appalling living conditions they endured. But as Glass (1968) and Steadman-Jones (1971) have shown, there was a persistent middle-class concern during the late nineteenth and early twentieth century regarding the concentration and segregation of the industrial working classes in the cities. Cooke-Taylor in 1842 commented on Manchester that:
As a stranger passes through the masses of human beings which have been accumulated around the mills and print-works in this and the neighbouring towns, he cannot contemplate those ‘crowded hives’ without feelings of anxiety and apprehension almost amounting to dismay. The population … is hourly increasing in breadth and strength. (Notes on a Tourin the Manufacturing Districts of Lancashire, quoted in Glass, 1968)
Sixty years later, Masterman (1904) used similar terms to describe middle-class reaction to the growth of British cities and the ‘dangerous classes’:
They dread the fermenting, in the populous cities, of some new, all powerful explosive, destined one day to shatter into ruin all their desirable social order. In these massed millions of an obscure life, but dimly understood and ever increasing in magnitude, they behold a danger to security and pleasant things.
Such fears focused on the rapidly growing northern industrial towns during the first half of the nineteenth century but, as Steadman-Jones (1971) noted in Outcast London:
In the period after 1850, fears about the consequences of urban existence and industrial society centred increasingly on London. For London, more than any other city, came to symbolize the problem of the ‘residuum’. (1971: 12)
These fears reflected both the size of London’s casual labour market (Green, 1995) and the huge size of the city itself. Victorian London was by far the largest city in the world, and during the course of the nineteenth century its population grew from just over one million to six and a half million. It also became increasingly segregated. Writing of London south of the river, Booth (1892) noted: ‘the population is found to be poorer, ring by ring, as the centre is approached, while at its centre there exists an impenetrable mass of poverty’
As Steadman-Jones (1971) clearly demonstrated, modern concerns regarding the so-called urban underclass closely parallel the concerns in the nineteenth century regarding urban ‘degeneration’, the ‘residuum’ and the ‘dangerous classes’. The behaviour of the lowest classes was thought to be vice-ridden, criminal and degenerate and it was widely feared that it was pathological and self-perpetuating. Ward (1989) gives similar evidence regarding the American city and quotes a report on the conditions of the poor in Boston in 1846 which could have been written of London. It noted that there was ‘a downward movement of the poorest classes … which if it not be checked, must sooner or later lead to a condition like that of the Old World where the separation of the rich and poor is so complete, that the former are almost afraid to visit the quarters most thickly peopled by the latter’ (Ward, 1989: 15).
Concern over urban disorder and poverty somewhat faded in the first half of the twentieth century. According to Glass, suburbanization may have helped to improve working-class living conditions, to tame working-class radicalism and soothe middle-class fears. But the point is that contemporary concerns over polarization and the rise of an urban underclass are not new. They have been around, in one form or another, for at least 150 years, rising and falling in prominence depending on changing circumstances (Beauregard, 1993a). The study of social segregation in nineteenth-century cities was an important issue in urban social geography and urban history in the 1970s and early 1980s (Dennis, 1980; Pooley, 1984), with a debate as to whether Victorian cities represented a new and distinctively modern form of social segregation, or whether they were a transitional form from pre-capitalist cities, to the modern capitalist city (Cannadine, 1977; Ward, 1975). Indeed, Ward (1980) has questioned the extent to which all Victorian cities conformed to the model of sharp class segregation suggested by Engels (see Doucet and Weaver, 1991; Harris, 1986; and Zunz, 1982 on residential class segregation in North America).
Contemporary Work on Class Structure and Residential Differentiation
Urban social geography in the 1960s and 1970s was characterized by the ever-more sophisticated quantitative analysis of urban residential structure, aided by the advent of computers and the publication of small area census data. Factorial and social area analyses of different cities proliferated in an attempt to link and test the traditional concentric (Burgess) and the sectoral (Hoyt) models of urban structure first formulated in the 1920s and 1930s. But, by the early 1970s, this type of work had reached its zenith (Johnston, 1970; Murdie, 1976). It was gradually realized that the ‘game of hunt the Chicago model’ (Robson, 1969) had run into an explanatory cul-de-sac, with rapidly decreasing returns to effort. More sophisticated quantitative approaches to the analysis of urban census data were producing less and less in the way of new understanding, and it was argued that a general theoretical approach to the study of urban residential structure linked to changes in class structure and other factors was needed (Harvey, 1975; Johnston, 1971; Timms, 1971).
The focus of research changed fundamentally in the mid-1970s and early 1980s with the publication of David Harvey’s (1973) Social Justice and the City, and his work on the residential structure of Baltimore (Harvey and Chatterjee, 1974). Along with Pahl’s (1970) work on urban managerialism, the nature of urban social geography and urban sociology shifted away from the study of residential patterns per se, and a concern with household choice and preference as the explanatory variables, towards a concern with the underlying economic and social processes which structured the nature of the urban housing market and, in combination with the existing class and ethnic structure, produced residential patterns. Attention shifted to study of the housing market and the production of urban residential space (Bassett and Short, 1980) and toward the analysis of constraints rather than choices and preferences (Short, 1978). By the mid-1980s, however, geographers had largely turned their backs on questions of segregation although, as discussed later, it is making something of a comeback in geography (Morrill, 1995; Peach, 1996) and urban sociology following Wilson’s (1987) seminal work, particularly in the context of the debate over the existence of an urban ‘underclass’ and its links to ethnicity.
Social Polarization and Residualization in the Housing Market
In recent years, the focus of attention has shifted from a focus on patterns of social and spatial segregation per se to the broader conception of polarization and duality, particularly in global or world cities (Friedmann and Wolff, 1982; Sassen, 1991; Mollenkopf and Castells, 1991). The idea of ‘polarization’ in urban social structure is not new. It was current in the early 1970s in London when there was a concern that the city was becoming polarized as a result of the outmigration of skilled manual groups to the New Towns and elsewhere, and a growing concentration of the highly skilled as a result of gentrification and of the less skilled because they were caught in a housing and employment trap (Cole, 1975; Eversley, 1972; Hamnett, 1976; Harris, 1973).
In addition, in Britain the decline of private renting in the inner city and the expansion of home ownership and council renting was also believed to be leading to a growing socio-tenurial polarization and to what was termed ‘residualization’ of council housing. In this interpretation, the socially mixed nature of private renting was giving way to a council house tenure increasingly dominated by the less skilled, the unemployed and economically inactive, and a variety of socially marginalized groups including single parent mothers and Afro-Caribbean tenants (Hamnett, 1984, 1987; Hamnett and Randolph, 1987; Henderson and Karn, 1984; Forrest and Murie, 1988). Home ownership, on the other hand, was becoming increasingly the preserve of professionals, managers and white-collar and skilled manual workers. This process was seen to be reinforced by the sale of better council houses in better areas to skilled and higher income tenants, leaving a rump council sector dominated by high-rise flats and poor housing on marginal estates (Forrest and Murie, 1988). This reinforced some existing tendencies towards the allocation of poor council housing to ethnic minorities (Parker and Dugmore, 1977). This research has been developed in the 1990s by Peach and Byron (1993) who have shown that Afro-Caribbean tenants in Britain are overrepresented in council housing and that Afro-Caribbean women are particularly concentrated in high-rise blocks. Similar processes operate in other European cities (Hegedus and Tosics, 1994; Kovacs, 1994; Musil, 1987; O’Loughlin and Friedrichs, 1996; Pichler-Milanovich, 1994; Sykora, 1994; Van Kempen, 1994) producing a variety of forms/patterns of residential segregation of disadvantaged groups.
Social Polarization and Duality in World Cities
The discussion of social polarization in the early and mid-1970s was largely concerned with measurement problems and techniques concerning the use of small area census data. The theoretical debate was relatively limited (Gordon and Harloe, 1991). It was not until the early 1980s that theoretical debate on polarization in world cities took off. The idea was popularized by Friedmann and Wolff (1982), who argued that whereas major cities have long had major inequalities in income and wealth, new processes relating to the international division of labour and production are at work in contemporary world cities which were concentrating both high level business services and a professional and managerial business elite.
At the other end of the spectrum, some cities, particularly in the third world, have seen the growth of a large informal, floating or street economy as well as the service economy needed to provide for the needs of the transnational elite. They suggest that the contrast between the business elite and the third of the population who make up the permanent underclass of the world city is striking, and they pointed to the ethnic dimensions of the underclass:
Many, though not all, of the underclass are of different ethnic origin than the ruling strata; often they have a different skin colour as well, or speak a different dialect or language. These immigrant workers give to many world cities a distinctly ‘third world’ aspect. There is a city that serves this underclass … Physically separated from and many times larger than the citadel of the ruling class, it is the ghetto of the poor. (Friedmann and Wolff, 1982: 323)
Thus, Friedmann and Wolff suggest that: ‘The primary fact about world city formation is the polarization of its social class divisions’ (1982: 322; added emphasis). But their use of the term polarization was undefined and simply implied there were sharp class divisions in world cities, which few observers would question. As I will argue, this lack of definition has characterized most of the subsequent work on polarization.
The concept of social polarization was taken up and developed by Saskia Sassen in a series of publications (1984, 1985, 1991). I have summarized Sassen’s argument at some length elsewhere (Hamnett, 1994). As part of her wider thesis about the role of global cities in the world economy, she argues that the structure of economic activity in global cities, particularly the dramatic growth of financial and business services and the decline of manufacturing industry, has: ‘brought about changes in the organization of work, reflected in a shift in the job supply and polarization in the income and occupational distribution of workers’ (Sassen, 1991: 9). She argues that this polarization in the occupational and income structure of global cities is a result of a number of interrelated processes. First she argues that the rise of business and financial services in global cities is leading to the creation of an occupational and income structure which contrasts with the occupational structure characteristic of manufacturing industry. This new occupational structure is comprised of a mixture of highly skilled and highly paid jobs and low-skill and low-paid jobs. As she puts it:
Major growth industries show a greater incidence of jobs at the high and low paying ends of the scale than do the older industries now in decline. Almost half the jobs in the producer services are lower income jobs, and half are in the two highest earnings classes. In contrast, a large share of manufacturing workers were in the middle earning jobs during the post-war period of high growth in these industries. (1991: 9)
In addition, Sassen argues that two other processes are generating increased polarization. These are, secondly, the secondary and derivative growth in low-skilled and low-paid jobs in hotels, catering, cleaning, personal services and the like, all of which are necessary to ‘service’ the new global service class. Thirdly, there is the growth of what Sassen calls a ‘downgraded’ manufacturing sector, characterized by a high concentration of informal and sweated low-skill and low-paid work. Sassen (1991) summarizes her thesis as follows:
new conditions of growth have contributed to elements of a new class alignment in global cities. The occupational structure of major growth industries characterized by the locational concentration of major growth centres in global cities in combination with the polarized occupational structure of the sectors has created and contributed to a growth of a high-income stratum and a low-income stratum of workers. It has done so directly through the organization of work and occupational structure of major growth sectors. And it has done so indirectly through the jobs needed to service the new high-income workers, both at work and at home, as well as the needs of an expanded low-wage work force. (1991: 9; emphases added)
This basic thesis has been consistently reiterated and elaborated by Sassen over the past 10 years and it is clear that, although she does not define social polarization precisely, it involves absolute growth of the occupational and income distribution at both the top and the bottom ends combined with an absolute decline in the middle. It is therefore seen to be more than a simple increase in income inequality. It is also clear that the growth of polarization, in Sassen’s view, is a product of changes in the social and spatial division of labour, which is seen to be particularly marked in global cities with their role as control and command centres and as centres of financial production. This interpretation of polarization has also received considerable support from the work of Harrison and Bluestone (1988) and Stan-bach (1979). It is linked to a growing concern about the so-called ‘disappearing middle’ (Kuttner, 1983; Lawrence, 1984; Levy, 1987), which was influential in the USA in the 1980s and still is. In addition, Sassen has linked the expansion of low-skill and low-pay jobs to the growth of the immigrant labour force who are attracted to global cities such as New York and Los Angeles by growing job opportunities. This is a consistent element of her thesis (Sassen, 1984, 1986, 1991). But, while this may be true of New York and Los Angeles (Clarke and McNicholas, 1996), it is very questionable to what extent polarization is characteristic of all global cities as Sassen suggests. Research on the Randstad, Holland (Hamnett, 1994b), Paris (Preteceille, 1995) and London (Hamnett, 1994a), suggests that these cities have not experienced occupational polarization. On the contrary, census and other data point to a consistent picture of upwards socio-economic shift. Table 10.1 illustrates this trend for London.
Pinch (1993) has noted, however, the debate over polarization is complex, and has taken different forms in the USA and the UK with Pahl (1988) taking a very different view of polarization from Sassen and Harrison and Bluestone. Pahl’s focus is on the division of work within and between households. It is not particularly urban, nor based on global cities. To this extent, the range of the thesis is quite different from that of Sassen. Pahl argues that a division is emerging between those ‘work-rich’ households who may have two or more members in employment, and ‘work-poor’ households whose members are unemployed and, because they lack the skills, contacts and income, are unable to engage in what Pahl terms ‘self-provisioning’. According to Pahl, there are a range of possibilities regarding the development of social polarization, ranging from the hour-glass structure which characterizes the United States, to the onion-shaped structure which he believes may be more characteristic of the UK. What is certain, is that modern cities, like their nineteenth-century forebears, possess sharp and growing differences in wealth and poverty (Dangschat, 1994; Hamnett and Cross, 1998 a and b; Haussermann and Sackman, 1994; Kesteloot, 1994; van Kempen, 1994). The key question, however, is whether these growing inequalities are a direct and unmediated product of economic restructuring, as Sassen suggests, or whether they may arise from differences and changes in state policies or other national differences (Hamnett, 1998).
| Table 10.1 Proportionate socio-economic change in London, 1961-91, economically active males (%) | ||||
| 1961 | 1971 | 1981 | 1991 | |
| Managerial | 11.9 | 13.9 | 16.1 | 20.3 |
| Professional | 4.8 | 6.0 | 6.4 | 8.2 |
| Intermediate and junior non manual | 23 | 23.5 | 22.0 | 23.7 |
| Skilled manual | 34.8 | 32.8 | 30.2 | 27.6 |
| Semi-skilled manual | 14.5 | 13.4 | 14.3 | 12.4 |
| Unskilled | 8.3 | 7.2 | 5.8 | 4.4 |
| Sub-total | 97.4 | 96.9 | l94.8 | 96.6 |
| Armed forces and occupation inadequately described | 2.6 | 3.1 | 5.2 | 3.4 |
| Total | 100 | 100 | 100 | 100 |
Source: Censuses of Population, 10% data
Dual Cities: Myth or Metaphor?
The 1980s have been seen to be an era of growing inequality between rich and poor which has been manifested particularly strongly in manor cities. As a consequence, we have seen a return to Disraeli’s nineteenth-century notion of the ‘two nations’, but this time in a specifically urban context. In their book Dual City, Mollenkopf and Castells (1991) comment that:
New York remains a capital for capital, resplendent with luxury consumption and high society … But New York also symbolizes urban decay, the scourges of crack, AIDS, and homelessness, and the rise of a new underclass. Wall Street may make New York one of the nerve centres of the global capitalist system, but this dominant position has a dark side in the ghettos and barrios where a growing population of poor people lives. (1991: 3)
This parallels the picture painted by Tom Wolfe in his novel The Bonfire of the Vanities (1988); but does this mean that New York is a ‘dual city’? Mollenkopf and Castells argue that specifically in terms of poverty and income inequality the answer is ‘yes’ and Castells (1993) argues, more generally, that:
the informational city is also the dual city … the informational economy has a structural tendency to generate a polarized occupational structure according to the informational capabilities of different social groups. Informational productivity at the top may lead to structural unemployment at the bottom or downgrading of the social conditions of manual labour. (1993: 254)
Following Sassen, Castells (1993) links also this process of dualization to immigration:
dualization of the urban social structure is reinforced when immigrant workers take on downgraded jobs. In a parallel movement, the age differential between an increasingly older native population in European cities and a younger population of newcomers and immigrants creates two extreme segments of citizens, polarized along lines of education, ethnicity, and age simultaneously. (1993:254-5)
Peter Marcuse (1989, 1993) has expressed serious doubts about the dual city thesis, arguing that the reality is more complex, and that the structure of the contemporary city is divided into several different groups and quarters, depending on the division of labour and on race and gender. Marcuse also argues that although the patterns have a spatial dimension, and their spatial characteristics influence their substance, they are: ‘not rigid spatial patterns in the old sense in which Burgess and Park tried to describe city structure’. He also strongly disputes any idea that contemporary social divisions are new in any fundamental sense: ‘The divisions of society, whether one chooses to speak of classes or socio-economic status or consumption or racial/ethnic groupings, are age-old: those derived from capitalism are hardly products of the postwar era’ (Marcuse, 1993: 357). He added that:
A divided city is certainly nothing new. Never mind the slave quarters of ancient Athens and Rome, the ghettos of the middle ages, the imperial quarters of colonial cities, or the merchant sections of the medieval trading cities. At least from the outset of the industrial revolution, cities have been divided in ways that are quite familiar to us. (1993: 354)
What is new, Marcuse suggested, is the extent of homelessness, the growth of gentrification and abandonment, the role of displacement as a mechanism of expansion by the middle classes, the growth of turf allegiance and battles, the role of government in promoting gentrification and the changing form of political cleavages, most of which stem from the nature of modern capitalism.
Social Polarization and Duality: All-Purpose Signifiers of Inequality?
While there is little doubt that the social and spatial structure of many major cities has become more sharply divided in recent years, and that urban income inequality has grown dramatically during the 1980s, there are three main problems surrounding the uncritical use of the term. First, there is a danger that the concept of polarization is used in a variety of undefined and often contradictory ways. Pahl (1988) and Pinch (1993) have pointed to the existence of quite different conceptions of polarization, and Marcuse (1989, 1993) has queried whether the notion of duality is appropriate or valid. Thus Mollenkopf and Castells (1991) in speaking of New York state that: ‘there is a process of social polarization, not just inequality: the rich are becoming richer and the poor are becoming poorer in absolute terms’ (1991: 401). But why should this be termed social polarization when an increase in absolute income inequality is the issue? Likewise, in an otherwise useful study of polarization and the crisis of the welfare state in Stockholm, Borgegård and Murdie (1994) discuss the changing position of Stockholm in the international economy, the changes in the welfare state in Sweden and impacts on income distribution, immigration and unemployment, all under the label social polarization, without anywhere defining what they mean by the term.
Indeed, the concept of social polarization is characterized more by shifting meanings than by precise definition. It has become an all-purpose general signifier of growing urban inequality and social division with the consequent disadvantages of ambiguity and lack of clarity. This issue is taken up below.
Secondly, there is the danger that the existence of social polarization is taken for granted rather than subjected to empirical analysis. Thus, polarization can easily become the received wisdom, the existence of which is simply assumed rather than problematized. While Dale and Bamford (1989) have empirically analysed aspects of Pahl’s view of social polarization, and Hamnett (1994b), Hamnett and Cross (1998a, 1998b) and Buck (1994) have examined the changing income and occupational structure of London in relation to Sassen’s thesis, the danger, as Fainstein et al. (1992) have perceptively pointed out, is that:
The images of a dual or polarized city are seductive, they promise to encapsulate the outcome of a wide variety of complex processes in a single, neat, and easily comprehensible phrase. Yet the hard evidence for such a sweeping and general conclusion regarding the outcome of economic restructuring and urban change is, at best, patchy and ambiguous. If the concept of the ‘dual’ or ‘polarizing’ city is of any real utility, it can serve only as a hypothesis, the prelude to empirical analysis, rather than as a conclusion which takes the existence of confirmatory evidence for granted. (1992:13)
Finally, there is a danger that the processes thought to be generating social polarization may be inadequately or incorrectly theorized or over-generalized and empirical evidence for certain forms of social polarization may be taken as proof of the validity of these process theories. I have previously criticized aspects of Sassen’s theory of social polarization and questioned whether her claim of the generality of the processes to all global cities is valid (Hamnett, 1994a, 1994b). Buck (1994) has also looked at some of the possible causes of the growing income polarization in London and New York which are not necessarily related to employment restructuring. It may be that certain forms of polarization are occurring, but not always for the reasons suggested by Sassen, Castells and others. As Levine (1992) puts it: ‘To what extent is economic restructuring a “single global process”? To what extent does the social polarization of cities flow inexorably from the intrinsic nature of post-fordist production or the inherent division of the service sector into high-wage and low-wage employment?’(1992: 175).
As I have argued elsewhere (Hamnett, 1994a, 1994b), there exist major doubts as to the extent that urban social polarization is a single process with similar causes and manifestations. This is not to suggest that the concept has no value. On the contrary, my concern is that unless the term is defined and used in a reasonably precise and systematic way it will lose its descriptive/explanatory power and become no more than a catch-all term. Beauregard (1993a and 1993b), however, suggests that the ambiguity and shifting signification of certain key terms is precisely the root of their importance.
The Rhetorical and Representational Role of Polarization
The ambiguity and shifting meaning of certain widely used concepts in social science has recently been discussed by Beauregard (1993b) in relation to representations of urban decline. The parallels with social polarization are close, and Beauregard’s views are worth detailed discussion. Virtually all his references to urban decline can be replaced by polarization. He states that:
Many of the notions that describe the social world are quite chaotic; they are concepts that bind together disparate behaviours and attributes. Commentors frequently use such terms without giving them specific meaning, knowing a focused definition would detract from their richness. They are not simple or rigid concepts and should not be treated that way. They are meant to elicit diverse impressions and resonate throughout the arguments in which they are used. Urban decline is one such concept (1993b: 188-9)
Beauregard is right in this respect and he is highly critical of approaches that seek to represent the notion of urban decline in terms of a range of objective indicators which produce ‘a typical social science, objectivist narrative’ which purports to represent the ‘real’ characteristics of urban decline via a series of tightly specified meanings and measurements. He argues that such narratives ‘negate what is most important about urban decline; its evaluative, emotional, and symbolic content’ and he states that the indeterminacy of such concepts stems from their cultural resonances rather than inadequate specification. He is strongly opposed to what he sees as the ‘reductionism’ implicit in trying to empirically specify concepts such as urban decline which do not give greater understanding. He views the purpose of such value-laden and representational concepts as being ‘mainly rhetorical’(1993b: 189)
I want to suggest that the concept of social polarization has many of the same characteristics as that of urban decline. It is inherently unstable, value-laden, symbolic and representational and its purpose is also mainly rhetorical. It is a multipurpose signifier for urban division and growing in quality. The same is true of the dual city idea. Indeed, Mollenkopf and Castells explicity point out that: ‘The dual city is a useful ideological notion because it aims to denounce inequality, exploitation, and oppression in cities, breaking with the organicist and technocratic views of cities as integrated social communities’ (1991: 405). But they also point out that its ‘underlying assumptions are rarely made explicit, because those who employ it tend to favour social critique over social theory. The political and emotional charge of a dualist approach and the failure to spell out its assumptions means that it cannot comprehend the complexity of urban social reality, which is certainly not reductible to a simple dichotomy’ (1991: 405). They note, however, that: ‘Even if the dual city metaphor can be scientifically misleading and often rhetorical or ideological, it nevertheless challenges us to explore the dimensions of growing inequality and explain the sources of the tendencies towards polarization’(1991: 11).
We therefore face something of a dilemma. While it may be politically useful to use concepts like social polarization because of their representational and rhetorical power, and implications of growing social divisions and inequality, such concepts can be empirically misleading and can divert attention away from what is happening to what we think is happening. While Beauregard raises an important issue, his anti-objectivist stance is problematic because it seems to fall victim to a rhetorical and critical trap no less powerful than the empirical one he so strongly rejects. If we do not spell out precisely what we mean and understand by social polarization we risk becoming slaves of an unexamined, imprecise or ill-defined concept. The fact that polarization is symbolic, value-laden and representational must not debar us from trying to deconstruct its uses and meanings.
Nor, having unpacked the various meanings and uses of the term, should we be debarred, as Beauregard seems to imply, from trying to assess whether the concept of polarization has any empirical validity. I disagree with his view that ‘objectivist narratives’ inherently negate what is most important about certain key concepts, namely their ‘evaluative, emotional and symbolic content’. Whilst objectivist narratives alone are insufficient and inadequate, so are purely rhetorical narratives. Social scientists should not be restricted to the construction and deconstruction of rhetorical narratives alone. While measurement and quantitative work does not, of itself aid understanding, to rely on theory or conceptual deconstruction alone, without trying to assess whether postulated processes have any empirical support, is to theoretically imprison us. Theoretical development should go hand in hand with empirical analysis. The two are not mutually antagonistic. But, and here I agree with Beaure-gard, the primary task is to unpick the chaotic concept of polarization to try to determine what its key elements are and how they relate together (if at all). Then, and only then, can we begin to investigate the form and nature of polarization. But, if we do not undertake empirical work, we remain solely in the realm of theoretical conjecture.
Unpacking the Concept of Social Polarization
Many commentators are convinced that we are witnessing a growth of social polarization in major capitalist cities. But the frequency with which the term is used is matched only by the general lack of a definition. Is it a synonym for growing social divisions and inequality, or does it mean something more precise and, if so, what? We must, I think, reject the notion that polarization can be simply used as a synonym for inequality, not least because inequality can take so many different forms (Gordon and Harloe, 1991: 383). Indeed, Esteban and Ray (1994), in a challenging, but extremely valuable, paper point to the major difference between inequality and polarization, showing that it is possible to have greater inequality without greater polarization and vice versa. Nor, in my view, should polarization be used to refer to increasing residential segregation by class, race, gender, etc., though this may certainly be related to growing social polarization at the city level. My reason for arguing this is that we already have a perfectly good term for this ‘segregation’, and we would then need to differentiate between social and spatial polarization which need not take place simultaneously.
It appears that Harloe and Gordon see polarization as a very specific form of inequality. Mollenkopf and Castells make a rather similar distinction, arguing that in New York: ‘there is a process of social polarization, not just inequality: the rich are becoming richer and the poor are becoming poorer’ (1991: 401). The distinction is also made by Kloosterman (1996), who suggests that while inequality in the distribution of earnings (or incomes) ‘refers to the extent of dispersion between given levels of earnings, polarization’ refers to the phenomenon of the disappearing middle, the shrinkage of the number of middle-income jobs (or income units) and a growth (absolute or relative) at both the top and bottom ends of the income distribution. This distinction is crucial because it points to what, for me, is the key element of polarization—a movement toward the poles of a given distribution. Thus, the Oxford English Dictionary defines polarization as: ‘an act of polarizing: the state of being polarized: development of poles: loosely, polarity’. This suggests that polarization can be conceived of either as a state or as a process. While it is quite possible to refer to a state or states of polarization, the term is commonly used to describe a process of polarization, where there is a movement towards the poles of the distribution. One of the best definitions is given by Marcuse (1989):
The best image is perhaps that of the egg and the hour glass: the population of the city is normally distributed like an egg, widest in the middle and tapering off at both ends; when it becomes polarized the middle is squeezed and the ends expand till it looks like an hourglass. The middle of the egg may be defined as ‘intermediate social strata’… Or if the polarization is between rich and poor, the middle of the egg refers to the middle-income group … The metaphor is not of structural dividing lines, but of a continuum along a single dimension, whose distribution is becoming increasingly bimodal. (1989: 699)
As Marcuse points out, polarization is a process whereby: ‘a distribution is becoming increasingly bi-modal’ irrespective of the precise dimensions along which polarization may be occurring. But polarization may be simultaneously taking place on a number of dimensions. Mollenkopf and Castells view polarization as being multi-dimensional, with distinct social and spatial dimensions:
the tendency towards cultural, economic, and political polarization in New York takes the form of a contrast between a comparatively cohesive core of professionals in the advanced corporate services and a disorganized periphery fragmented by race, ethnicity, gender, occupational and industrial location, and the spaces they occupy. (1991: 402)
We need to specify whether we are speaking of employment, occupation or income, and whether polarization is relative or absolute. This is important because polarization may be occurring in certain respects but not in others, and the causes may be quite different.
Polarization of the Occupational and Income Structure
In terms of employment, polarization can be used to refer to an increase in the number (or proportion) of the highly skilled and the low-skilled with a decline in the number (or proportion) of the middle groups. This is the sense in which Sassen uses the term, and she links it to an increase in the number of highly paid and low-paid workers and to a decline in the number of middle-income workers, both of which result from the shift from manufacturing to financial and business services which is seen as particularly marked in global cities. Mollenkopf and Castells also suggest that the dual city notion usefully emphasizes one trend—both the upper and the lower strata of a given society grow at disproportionate rates. Thus, in the perspective of the polarization thesis, the dual city becomes a simple [sic] matter of empirical testing of two basic questions:
Are the top and bottom of the social scale in a given city growing faster than the middle (with the key methodological issue being how to construct a scale to measure social distribution)?
How does such polarization, if it exists, translate into spatial distribution at the top and bottom of the local society, and how does such specific residential location affect overall socio-spatial dynamics? (1991: 407)
Mollenkopf and Castells are very clear about the outcome of the polarization process and how it should be measured, although they do not make clear whether they are speaking of absolute or relative change. But the argument about the changing size of the rich and poor groups is, however, very different from the argument that there is a widening gap between the average incomes of the rich and those of the poor: that the rich are becoming richer and the poor, poorer (either relatively or absolutely). Both theses may be correct, but the causes may be very different and it is not legitimate to use evidence for one to support claims about the existence of the other.
This may seem mere definitional nit-picking, but it has major implications for attempts to empirically determine whether polarization is (or is not) occurring in certain, specific, forms (see Buck, 1994). It is important to note here that although Sassen and Castells have argued that occupational and income polarization go hand in hand, this is not necessarily the case. I have pointed out that, in the Randstad, Holland and London (Hamnett, 1994a, 1994b) there is strong evidence of a widening income gap between rich and poor but no evidence of occupational polarization. This may reflect the fact that, in the Netherlands, the official figures are based on the employed and exclude the unemployed, but Sassen’s thesis relates to the changing structure of employment (see Burgers, 1996). Similarly, Prete-ceille (1995) found that in Paris, the occupational evidence points to professionalization rather than to polarization. As Kloosterman points out, analytically:
two concepts of polarization in urban areas can be distinguished. The first one covers the whole city population and includes, therefore, those people without work … The second concept of urban polarization is more modest and deals only with those that have paid work. (1994: 2)
More generally, however, it appears that growing income inequality is not necessarily accompanied by a polarization of occupational structure as Sassen and Castells maintain. As Hamnett (1994a) has argued, there is evidence that, rather than polarizing, the occupational structure of Western capitalist societies appears to becoming more professionalized (Wright and Martin, 1987). This analysis is supported by Esping-Andersen (1993), who argues that occupational upgrading is inherent in the post-industrial trajectory in the United States. Indeed, Kloosterman suggests that:
During the 1970s and 80s, a decoupling seems to have taken place between the occupational level and the wage level in the United States … According to Esping-Andersen, a polarization of the occupational structure has been accompanied in the US by a polarization of wage structure. (1996: 468)
A similar shift appears to have occurred in Britain, in that the socio-economic structure has shifted upward whilst earnings and income inequality has risen considerably (Hamnett and Cross, 1998a, 1998b). But, this is not the result of an increase in the number of the less skilled and low paid as Sassen suggests, but from the impact of rising professional and managerial incomes, massive tax breaks for the rich (Hamnett, 1994a), growing unemployment and small increases in rates of government assistance for the unemployed or low paid. The key question then becomes what factors are leading to occupational depolarization and income polarization and how is the existence of the two processes to be explained and linked together? It is also necessary to examine the extent to which occupational depolarization and income polarization may, or may not, be characteristic of other Western countries, and what factors may be leading to the existence of different tendencies. Differences in welfare state regimes may be very important.
The Underclass Debate
The notion of the ‘residuum’, a category of the poor and unemployed outside or beneath the rest of the social structure, was common in nineteenth-century London, as Steadman-Jones (1971) has shown. The existence and growth of this group was often linked to what was termed ‘urban degeneration’. But, as Jones shows, the primary cause was the casualization of parts of the London labour market, and the creation of a large group of structurally unemployed or underemployed. This process of pauperization has also been discussed by Green (1995). Over the past hundred years, similar concerns have resurfaced from time to time, usually, though not always, voiced by right-wing commentators. Smith (1992) suggests that: ‘the idea of an underclass is a counterpart to the idea of social classes, and acquires its meaning within that same framework’. He thus argues that the underclass, if they exist, are those who fall outside the standard class schemas in that they belong to family units who have no stable relationship with the mode of production. Indeed, Marx himself used the term ‘lumpenproletariat’. Runciman argues for a similar definition of the underclass: ‘those members of … society whose roles place them more or less permanently at the economic level where benefits are paid by the state to those unable to participate in the labour market at all’ (1990: 388).
In the past 10 years the ‘underclass’ has become a common term in academic and media discourse. It first appeared in 1962 when Gunnar Myrdal used the term as ‘a purely economic concept, to describe the chronically unemployed, underemployed, and unemployables being created by what we now call the post-industrial economy’ (Gans, 1990: 271). But, as Gans (1990, 1993) and others (Robinson and Gregson, 1992; Wilson, 1987) have pointed out, the concept was subsequently hijacked by the radical right who employed it to encapsulate the idea of a culture of poverty, family dissolution and criminality in the inner cities, particularly in the black community.
In the 1960s, Oscar Lewis’s (1968) anthropological concept of a culture of poverty became popular, and in the 1970s, Edward Banfield (1974) popularized the view that the problem of what he termed the ‘lower class’ (below the working class) was primarily the result of a pathological transmitted culture of present orientation and low expectations. He stated that:
the lower class individual lives from moment to moment. If he [sic] has any awareness of a future, it is of something fixed, fated, beyond his control; things happen to him, he does not make them happen. Impulse governs his behaviour, either because he cannot discipline himself to sacrifice a present for a future satisfaction or because he has no sense of the future. He is therefore radically improvident: whatever he cannot use immediately he considers valueless. His bodily needs (especially for sex) and his taste for ‘action’ take precedence over everything else—and certainly over any work routine. (1974: 61)
The lower-class household is usually female-based. The woman who heads it is likely to have a succession of mates who contribute intermittently to its support but take little or no part in rearing the children. In managing children, the mother is characteristically impulsive: once children have passed babyhood they are likely to be neglected or abused, and at best they never know what to expect next. A boy raised in such a household is likely at any age to join a corner gang of other such boys and to learn from the gang the ‘tough’ style of the lower-class man. (1974: 62)
He concluded that:
So long as the city contains a sizeable lower class, nothing basic can be done about its most serious problems. Good jobs may be offered to all, but some will remain chronically unemployed. Slums maybe demolished, but if the housing that replaces them is occupied by the lower class it will shortly be turned into new slums. Welfare payments may be doubled or tripled and a negative income tax instituted, but some persons will continue to live in squalor and misery. New schools may be built, new curricula devised, and the teacher-pupil ratio cut in half, but if the children who attend these schools come from lower class homes, the schools will be turned into blackboard jungles, and those who graduate or drop out from them will, in most cases, be functionally illiterate. (1974:234)
Banfield’s radical cultural pessimism affords no solution but that of selective eugenics and birth control measures for the lower classes. He is of the view that welfare measures and other ‘good works’ are doomed to failure in that the lower classes are incapable of taking advantage of measures put forward for the improvement of their lot. A culture of low expectations and present orientation undoubtedly exists in some groups, but what Banfield critically fails to analyse are the causes and reasons why it has come into existence or grown in significance. Cultures do not simply spring into being of their own volition. As Banfield admits: ‘From the beginning, the cities of the United States have had upper, middle, working and lower classes … [but] [T]he relative strength of the various classes have varied greatly from time to time and place to place’ (1974: 63).
The question of why the relative strength of various classes and cultures has varied from time to time and place to place, and their causes, is not discussed. According to Katz (1993) the debate on the underclass in the USA accelerated in 1977 when Time Magazine announced the emergence of a menacing new underclass in America’s inner cities. Drugs, crime, teenage pregnancy and high unemployment, not poverty, defined the ‘underclass’, most of whose members were young and from ethnic minority groups (see also Devine and Wright, 1993).
Behind the [ghetto’s] crumbling walls lives a large group of people who are more intractable, more socially alien and more hostile than almost anyone had imagined. They are the unreachables: the American underclass … Their bleak environment nurtures values that are often at odds with those of the majority—even the majority of the poor. Thus the underclass produces a disproportionate number of the nation’s juvenile delinquents, school drop-outs, drug addicts and welfare mothers, and much of the adult crime, family disruption, urban decay and demand for social expenditures. (Time Magazine, 1977)
Katz states that with the publication of Auletta’s (1982) book The Underclass, the term ‘secured its dominance in the vocabulary of inner-city pathology’. For Auletta, the underclass was a relatively permanent minority among the poor who fell into four distinct categories, which he defined as (a) the passive poor, usually long-term welfare recipients; (b) the hostile street criminals who terrorize most cities, and who are often school dropouts and drug addicts; (c) the hustlers, who, like street criminals, may not be poor and who earn their livelihood in an underground economy; and (d) the traumatized drunks, drifters, homeless shopping-bag ladies and released mental patients who frequently roam or collapse on city streets (1982: xvi).
This right-wing portrayal of the underclass as a group of welfare-dependent, demoralized and behaviourally deviant individuals, frequently black and from single-parent families in the inner city, achieved further prominence with the publication of Charles Murray’s (1984) Losing Ground. Murray argued that the principal cause of the growth of the underclass was the welfare programmes which had eroded the will to work and the incentives for stable family life. The right wing consistently stress culture and behaviour as the key attributes and causes of the emergence of an ‘underclass’, but even distinguished black analysts such as WJ. Wilson (1987) accept the term, arguing that the liberal perspective on the ghetto underclass has become less persuasive, primarily because many liberal commentators ‘have been reluctant to discuss openly or… even to acknowledge the sharp increase in social pathologies in ghetto communities’ (1987: 6). As a result, he argues that conservatives have effectively defined and dominated the debate in recent years. But, says Wilson:
Regardless of which term is used, one cannot deny that there is a heterogeneous grouping of inner city families and individuals whose behaviour contrasts sharply with that of mainstream America. The challenge is not only to explain why this is so, but also to explain why behaviour patterns in the inner city today differ so markedly from those of only three or four decades ago (1987: 7)
Included in this group are individuals who lack training and skills and either experience long-term unemployment or are not members of the labour force, individuals who are engaged in street-crime and other forms of aberrant behaviour, and families that experience long-term spells of poverty and/or welfare dependency. These are the populations to which I refer when I speak of the underclass. I use this term to depict a reality not captured in the more standard designation lower-class. (1987: 8)
Notwithstanding Wilson’s view that it is crucial for liberals to face up to the realities of life in the black inner city ghettos (vividly depicted in films such as Boyz ‘n the Hood), Fainstein (1993) argues (pace Beauregard), that the idea of the underclass has become the dominant discourse of race and class in America, and that we need to break free of the discourse if we are to make progress, both analytically and in political and policy terms. He argues that: ‘Whatever their political and theoretical perspective, participants in the discourse of the underclass share a deep narrative. Like other deep narratives, that of the underclass is both explicit and implicit, saying much in its omissions’ (1993: 385). Fainstein states that its logic is relatively transparent, constructed along four lines. He summarizes these as follows: underclass terminology offers a way of speaking about race in a language of class that implicity rejects the importance of race; research on the underclass tends to study the attributes or behaviours of a category of the population that is nominally separated from other groups and from processes that affect larger populations. As a result of the problematic and value-laden nature of the term, Robinson and Gregson (1992) suggest in their useful review paper that the negative connotations of the term are such that it is best not to use it, and Mingione (1996) is also sceptical although the term now seems to have acquired a life of its own which authors feel compelled to address even though they reject its validity (Lee, 1994). Wilson (1991) now substitutes the term ‘ghetto poor’ in the USA and makes the point that although there is a distinctive culture it is largely a response to structural changes and constraints (see Holloway, 1990). One problem of the growing use of the term in Europe is the extent to which it is valid to utilize concepts derived from one social context and apply them to another (Martiniello, 1996; Musterd, 1994; Wacquant, 1993). Murray (1990 has suggested that there is an emergent British underclass, but Lydia Morris (1993) in her work on Hartlepool concludes that whilst there are differences in kinship and friendship networks between the unemployed, the securely employed and the insecurely employed, they are differences of degree rather than kind, which although they add to the disadvantages of the unemployed, provide no evidence of any ‘distinctive’ underclass culture. She concludes that the notion of the ‘underclass’ is:
an oversimplification, contaminated by its use as a tool of political rhetoric, which has been too readily applied to complex social phenomena. Its use is rarely supported by empirical research of the kind necessary to substantiate the varying claims with which it is associated. (1993: 411)
Buck (1996) argues that the difficulties with the term suggest that we should be very cautious in using it unless we can specify coherently what it means. He states that: ‘The reason for using the term “underclass” is that it carries an implication that the group experiencing social and economic marginalization is … characterized by homogeneity, stability and social segregation’ (1996: 279). But he argues that although there has been a large increase in the number of households experiencing long-term inactivity, and hence poverty, in Britain, that the group displays considerable heterogeneity and there is no evidence of high levels of spatial segregation. Thus, he concludes that although it is difficult to draw firm comparative conclusions:
comparison with the ideal type model of urban poverty in the USA suggests that differences in state welfare policy and in labour market regulation may lead to poverty and marginality being constituted in very different ways … we need to be very cautious in translating definitions of marginality and exclusion from one society to another. (1996: 297)
Unfortunately, the term has been taken over by the mass media, who employ it as a handy label without systematically examining the extent to which the term possesses analytical or empirical validity. Consequently, it is frequently reproduced willy-nilly despite the efforts of authors such as Loic Wacquant (1993, 1996a, 1996b) to analyse the term critically and the ideologies which underlie it. To this extent, the notion of an underclass as a group of people outside the economic mainstream by reason of chronic unemployment and structural change in the economy, has escaped its initial authors, aided and abetted by the new right who have a strong interest in promoting the idea of a behavioural culture of poverty which is reinforced by the evils of state aid. It seems clear that the idea of the underclass is not simply going to go away merely because the centre-left disapprove of its connotations and the ways in which it is used to support a punitive, anti-welfare agenda. As Wilson points out, it is necessary to go on the intellectual offensive and show that the term is value-laden and empirically suspect. Alternatively, if it is thought that there are groups of people who are (semi-) permanently excluded from the labour market and in more or less permanent poverty, it is necessary to show what the consequences of this are, socially, culturally and economically, rather than uncritically reproducing the idea of the underclass by conceptual repetition. The same applies both to the notion of social polarization and the ‘ghetto’ as Wacquant (1993,1996a, 1997) has very powerfully argued. None the less, the term is continually reproduced via the mass media as a convenient hook. As I write, an article by John Lloyd (1996) appeared in the New Statesman on Labour’s plans in Britain to combat social exclusion. Lloyd states that one of the New Labour thinkers, Geoff Mulgan, ‘insists on the nomenclature of “social exclusion” as against the “poor” or even the newer “underclass”’. Lloyd states that this is because poverty is only one attribute of those at the bottom of the heap: they are more properly defined as excluded because they live outside the worlds of work, education and sociability. The article was termed, A plan to abolish the underclass’.