is sydney a divided city ethnically

22
356 Australian Geographical Studies November 2004 42(3):356–377 Is Sydney a Divided City Ethnically? MICHAEL POULSEN 1 , RON JOHNSTON 2 and JAMES FORREST 1 , 1 Macquarie University, Australia ; 2 University of Bristol, UK Abstract Recent claims that Sydney’s ethnic residential concentrations are a permanent feature and that the city is rapidly turning into a city bifurcated along an ethnic divide cannot be sustained by the evidence. An understanding of multicultural policies as they operate in Australia, and of segregation as essentially a transi- tory phenomenon there, suggests that social bifurcation is unlikely to occur. Immigrant concentrations in the poor suburbs are nothing new, and do not imply the entrapment of ethnic groups. While immigrant residential concentrations are increasing in size, commensurate with the influx of large numbers of migrants from non-English speaking backgrounds over the past 30 years, the evidence suggests that claims of bifurcation are a product of the usage of birthplace data, the aggregation of ethnic groups into just two groups and a failure to take account of the degree of mixing. Evidence from both birthplace and ancestry data using different levels of ethnic aggregation and more appropriate analyses shows that Sydney is dominated by an intermixing of different ethnic groups with each other and with the host society, and not by high levels of ethnic segregation. Ancestry data from the 2001 Census show the spatial assimilation into the host society of the grandchildren, but less so the children, of immigrants in Sydney with similar rates for both the white non-Anglo-Celtic and the Asian populations. KEY WORDS ethnic concentrations; multiculturalism; immigration; spatial assimilation; Sydney; ethnic mix Introduction It is 30 years since multiculturalism was embraced as official policy regarding the absorption of immigrant groups into Australian society. Dur- ing that time, the ethnic mix of Australian society, and especially that of its largest city, Sydney, has changed dramatically (Birrell and Rapson, 2002). Previously, Sydney, like the rest of Aus- tralia, was almost exclusively European in its population’s ethnic origins. Approximately 87% of Australia’s population was Anglo-Celtic in origin in 1971 (Jupp, 1989). The majority of the non Anglo-Celtic ethnic population comprised southern, central and eastern Europeans, Leba- nese, Anglo-Indians and New Zealand Maori who had arrived in Australia in the 1950s, 1960s and 1970s. With the removal of any ethnic crite- ria from the Immigration Act in 1973 following the ending of the White Australia policy in 1966, new immigration flows increasingly included a strong Asian element. Concurrently, a ‘multi- cultural’ policy was developed (Williams and

Upload: mq

Post on 23-Jan-2023

1 views

Category:

Documents


0 download

TRANSCRIPT

356

Australian Geographical Studies

November 2004

42(3):356–377

Is Sydney a Divided City Ethnically?

MICHAEL POULSEN

1

, RON JOHNSTON

2

and

JAMES FORREST

1

,

1

Macquarie University, Australia

;

2

University of Bristol, UK

Abstract

Recent claims that Sydney’s ethnic residential concentrations are a permanentfeature and that the city is rapidly turning into a city bifurcated along an ethnicdivide cannot be sustained by the evidence. An understanding of multiculturalpolicies as they operate in Australia, and of segregation as essentially a transi-tory phenomenon there, suggests that social bifurcation is unlikely to occur.Immigrant concentrations in the poor suburbs are nothing new, and do not implythe entrapment of ethnic groups. While immigrant residential concentrations areincreasing in size, commensurate with the influx of large numbers of migrantsfrom non-English speaking backgrounds over the past 30 years, the evidencesuggests that claims of bifurcation are a product of the usage of birthplace data,the aggregation of ethnic groups into just two groups and a failure to takeaccount of the degree of mixing. Evidence from both birthplace and ancestrydata using different levels of ethnic aggregation and more appropriate analysesshows that Sydney is dominated by an intermixing of different ethnic groupswith each other and with the host society, and not by high levels of ethnicsegregation. Ancestry data from the 2001 Census show the spatial assimilationinto the host society of the grandchildren, but less so the children, of immigrantsin Sydney with similar rates for both the white non-Anglo-Celtic and the Asianpopulations.

KEY WORDS

ethnic concentrations; multiculturalism; immigration; spatialassimilation; Sydney; ethnic mix

Introduction

It is 30 years since multiculturalism was embracedas official policy regarding the absorption ofimmigrant groups into Australian society. Dur-ing that time, the ethnic mix of Australian society,and especially that of its largest city, Sydney,has changed dramatically (Birrell and Rapson,2002). Previously, Sydney, like the rest of Aus-tralia, was almost exclusively European in itspopulation’s ethnic origins. Approximately 87%of Australia’s population was Anglo-Celtic in

origin in 1971 (Jupp, 1989). The majority of thenon Anglo-Celtic ethnic population comprisedsouthern, central and eastern Europeans, Leba-nese, Anglo-Indians and New Zealand Maoriwho had arrived in Australia in the 1950s, 1960sand 1970s. With the removal of any ethnic crite-ria from the Immigration Act in 1973 followingthe ending of the White Australia policy in 1966,new immigration flows increasingly includeda strong Asian element. Concurrently, a ‘multi-cultural’ policy was developed (Williams and

Is Sydney a Divided City Ethnically? 357

© Institute of Australian Geographers 2004

Batrouney, 1998, 261). This provided for theright of individual ethnic groups to retain andexpress their cultural identities, and for socialjustice and equal opportunity, but within aframework of the full utilisation of the economicpotential of all Australians (Smolicz, 1995).

From the late 1940s, the social compositionof Australia’s population has been transformed.Through the 1950s and 1960s into the 1970s, acentury and a half of Anglo-Irish sourced migra-tion changed to incorporate a significant Euro-pean component. Then, with the dismantling ofthe White Australia policy during the 1970s, Asiabecame a major source of immigrants. Many ofthese people became concentrated in less skilled,manufacturing industries where they often be-came trapped by:

Lack of English language proficiency . . .lack of transferable work skills due to the[largely] unskilled nature of employmentgained, lack of education or qualifications,or underevaluation (in an Australian context),of such skills, education or qualifications asthey [had]. (Morrissey

et al

., 1992, 15)

Residential segregation in less well off areas isa consequence of this labour market situation(Johnston

et al

., 2001), reinforced by othersocial considerations such as choice among theimmigrant groups themselves (Zang and Has-san, 1996). As Massey (1985) has argued, suchethnic concentrations continue to expand as longas immigration rates are in excess of the levelsof spatial assimilation which, in turn, dependson ‘normalisation’ of access to the labour mar-ket as problems with proficiency in English,deficiencies in education and lack of recognitionof skills are overcome (Iredale, 1997, 98). Immi-grants from non-English speaking backgrounds,whether from Europe prior to the ending of theWhite Australia policy, or subsequently fromAsia and the Middle East in particular, are there-fore behaving little differently from the Anglo-Celtic immigrants of an earlier period.

Sydney has been recognised through interna-tional comparative research on segregationlevels to be an exemplar of ethnic mix (Poulsen

et al

., 2002a). It does not have the ghettos whichcharacterise many American cities, nor does ithave the large, polarised ethnic enclaves whichcharacterise many New Zealand cities. It is acity of mix, where the majority of immigrantsand their descendants live in shared spaces withthe host community. Healy and Birrell (2003)counter these views with the claim that theincreasing concentration of non-English speak-ing background (NESB) immigrants ‘challengesthose who argue that such concentrations wouldbe relatively transitory (following the pattern ofearlier post-war migrant groups) and not a moreenduring feature of contemporary metropolitandevelopment in Sydney and Melbourne’ (2003,69). They conclude that, in Sydney:

. . . not only is this city’s population growing,it is also bifurcating . . . there are now twoSydneys — one increasingly dominated bylow to moderate-income non-English-speakingmigrant communities in the West and South-West and the other comprised of establishedinner affluent areas and predominantly English-speaking ‘aspirational’ areas in the metropol-itan periphery. (Healy and Birrell, 2003, 65)

While these claims are provocative, Healy andBirrell’s analyses cannot be seen to substantiatetheir conclusions. There are four key problemswith their research. First, in limiting theiranalyses to a 2001 Census internal migrationmatrix based on birthplace data, they haveignored the subsumption of descendents of theNESB immigrants within the Australian-bornpopulation. Without using ancestry data we donot know whether the Australian born leavingthe poor NESB Local Government Areas areAnglo-Celtic or the descendants of the NESBimmigrants. Secondly, they do not split the NESBpopulation into those who were primarily earlierpost war immigrants and more recent immigrants,despite Healy and Birrell’s claim that earlierpost war immigrants display the expected pat-tern of spatial assimilation. Thirdly, there is noassessment of the levels of spatial assimilation— reflecting social assimilation to the pointwhere socio-economic status (income, occupation,

358 Australian Geographical Studies

© Institute of Australian Geographers 2004

education), not ethnicity, determines the life chan-ces achieved by the children and grandchildrenof the NESB immigrants. Fourthly, they use anaggregation level that is not common in inter-national ethnic research. They split the populationinto Australians and NESB immigrants. In otherwords, they substitute the study of ethnic com-munities with a study of all NESB immigrants.

What makes Healy and Birrell’s researchcontribution significant is their implication thatthe Australian government’s immigration policysettings are wrong and that the aims of multicul-turalism are not being advanced. Furthermore,while not calling it ‘white flight’, they implythat the Australian born population is fleeingimmigrant areas as they expand. As such, thepolicy implications of their research are that thegovernment’s immigration levels should bereduced; otherwise Sydney faces a social dis-aster as a result of the bifurcation of its societybetween NESB immigrants and the Australianborn. So the question that underlies our researchis, are Healey and Birrell correct in either orboth of their diagnosis and their prognosis?

Support for the case for mix is provided by anumber of other studies. Burnley (1994) usedthe 1986 ancestry data to focus on the post warimmigration of non-Anglo-Celtics under the‘White Australia’ policy, primarily those fromGreece, Italy and Lebanon. He assessed thenature of spatial assimilation of these ethnicgroups and concluded that immigrant communi-ties effectively lasted for one generation inpoorer areas of the city and that the second gen-eration (their children) moved out and foundedsecondary concentrations in middle and outersuburbia. For the third generation (their grand-children), he found no evidence of ethnic con-centrations, because they had become spatiallyassimilated with the host society. Jones (1996)similarly argued that ethnic concentrations woulddissolve as a result of spatial assimilation con-sequent upon the full integration of migrantgroups into the workforce. Working with birth-place data, Poulsen

et al

. (2001a) argue thatethnic groups in Australian cities are among theleast segregated in the English-speaking world,

and that ethnic ghettos (as normally understood)do not exist there. Instead, they view Australiancities as best characterised by their degree ofethno-cultural mix rather than segregation, andargue that the host community, too, is increas-ingly becoming less segregated spatially.

Using two new approaches to the analysis ofmigrant group segregation and desegregation,and ancestry data from the 2001 census, thisstudy seeks to resolve these issues. Theseapproaches — unlike those usually employed instudies of residential segregation — focus onabsolute levels of separation, which are muchmore suited to evaluating Healy and Birrell’sarguments, as indicated below. By using them,we show that reliance on birthplace data aloneunderstates the amount of residential separationof ethnic groups in Sydney but that, even whenusing ancestry data, there is no evidence ofSydney being an ethnically polarised city.

Immigration and multiculturalism

Australia’s immigration policy is aimed prima-rily at a mix of skilled and educated migrantsregardless of country or race of origin throughtwo programmes.

1

The skills program pointstest, which migrants must pass contains thepreconditions for maximising the spatial assimila-tion of migrants through its weightings on skilledoccupations required in Australia, and on educationlevels, age, English language ability, Australianqualifications, and spouse skills. The humanitarianprogram effectively avoids these checks, but thisonly accounts for approximately 10% of the totalimmigration programme (12 000 of the 110 000places in 2003/4). The Australian Government’spolicy view of multiculturalism is that it:

. . . recognises and celebrates Australia’scultural diversity. It accepts and respects theright of all Australians to express and sharetheir individual cultural heritage within anoverriding commitment to Australia and thebasic structures and values of Australiandemocracy. It also refers specifically to thestrategies, policies and programs that aredesigned to:

Is Sydney a Divided City Ethnically? 359

© Institute of Australian Geographers 2004

• make our administrative, social and eco-nomic infrastructure more responsive tothe rights, obligations and needs of ourculturally-diverse population;

• promote social harmony among the differ-ent cultural groups in our society; and

• optimise the benefits of our cultural diver-sity for all Australians.

(DIMIA, 2003)

Multiculturalism in contemporary Australia isprimarily targeted at the first generation ofimmigrants, leaving the education system andthe media to complete the integration of theirchildren and grandchildren into the wider soci-ety (Beckett, 2001; Al-Haj, 2002). However, theattitude of many Australians to multiculturalismis, at best, ambivalent, linked as it is to issues ofcultural privilege and national identity (Johnson,2002; Dixson, 1999). In 1999, a National Multi-cultural Advisory Council, appointed by PrimeMinister John Howard, argued that:

The British and Irish heritage, which includesour democratic system and institutions, ourlaw, the English language, much of our humourand oft-quoted distinctive values of the fair go,egalitarianism and mateship, together providethe foundation on which

Australian multicul-turalism

[italics added] has been built. (NMAC,1999, 4)

The media, with the exception of one State-sponsored, multicultural television channel, con-tinue to reinforce the ‘white’ Australian culturethrough their under-representation and misrepre-sentation of ethnic minorities (Ley and Murphy,2001). The preferential position of the hostsociety remains, through its culture, language,institutions and laws (Marden and Mercer, 1998).

Compared with these perspectives, officialstances on multiculturalism are overtly support-ive, and have been integrated into the legal sys-tems of the various states. In New South Wales,for example, a multiculturalism act specifies that‘respect and provision for culture, language andreligion of others [occurs] within an Australianlegal and institutional framework where English

is the common language’ and that for ‘all peoplein a multicultural society . . . there is a unifyingcommitment to Australia, its interests andfuture’ (New South Wales Community RelationsCommission and Principles of MulticulturalismAct 2000). Many western nations, though notthe United States, have their own variants ofthis policy (MacLaughlin, 1998). As in most ofthese, the Australian government’s multiculturalpolicy is implemented through myriad clausesand amendments to a variety of Acts, such as theCrimes Act and the Family Law Act. It is thislegal framework that makes it different fromthe assimilationist policy in the United States,where assimilation is the expectation, based ona belief in the superiority of the American cul-ture over all other cultures (Bannerji, 2000).

Immigration and ethnic concentrations

Ethnic residential concentrations are the resultof either discrimination or a group’s need forcommunity support. Such support takes theform of the development of a community infra-structure, which consists of businesses, religiousbuildings, social and sporting clubs, and supportnetworks. These ethnic concentrations candevelop quickly, then either disappear after afew decades or continue to exist for generations.Ethnic concentrations dissolve through the proc-ess of spatial assimilation, that is movement outof those areas and mixture with the host society.The duration of immigrant ethnic residentialconcentrations, according to Massey (1985), isdependent on immigration policy, in that thecontinuing sourcing of immigrants for a groupoffsets the degree of spatial assimilation that isoccurring. In general, spatial assimilation seesthe less socially distant immigrants, and thechildren of those immigrants when they estab-lish their own households moving to suburbsnearby so that they can live in close proximityto parents and their immigrant community. Theirchildren, the grandchildren of the immigrants,usually scatter across the city, as their links tothe original host community break down. Theexceptions to this process of spatial assimilationare those groups which, because of social distance

360 Australian Geographical Studies

© Institute of Australian Geographers 2004

from either their host society or another ethnicgroup, are discriminated against (e.g., AfricanAmericans in America), those who, because ofreligious ties, need to retain their close proxim-ity to their religious institutions (e.g., the Jewishpopulation), and those who, for cultural reasons,prefer to continue to live together (e.g., the NewZealand Maori) (see Khoo and Birrell, 2002;Burnley, 1994; for a comment on the US expe-rience, and delayed, even ‘downward’ assimila-tion, see Portes and Rumbaut, 2001, 44). HenceJamrozik

et al

. (1995, 110) refer to multiculturalismin Australia as ‘assimilation-in-slow-motion’, ora delayed assimilation model.

In the Australian context, spatial assimilationis an ongoing process for most ethnic groups;hence immigration policy is the key elementin the duration of ethnic concentrations. Whereimmigration rates remain constant, spatial as-similation and immigration will, after approxi-mately 30 years, reach an equilibrium state andthe ethnic concentrations will cease to expand(Figure 1, type 1). If immigration levels declinerelative to spatial assimilation rates, or rates ofspatial assimilation are increased, then ethnicconcentrations will start to contract (type 2).However, contraction may not continue until allethnic concentrations are dissolved; some ethnicareas may remain, to represent a pluralist out-come with some cultural groups (or a substantialnumber of their members) retaining their sepa-rate identity. In such a case some ethnic concen-trations will continue to expand indefinitelyas long as the source and levels of that ethnicgroup’s immigration are maintained. Alterna-

tively, if the immigration rate increases or if spa-tial assimilation is unable to achieve a rate equalto that of immigration, then the size and spatialextent of the ethnic concentrations will alsocontinue to increase (type 3). Levels of concen-tration are also dependent on the degree of mixwithin the host community. As that degree ofmix increases, the differences between areas ofethnic concentration and the host communitydecline as both tend to merge into each other.Potentially it is possible for ethnic concentra-tions to cease to be important if mix and hybrid-ity become characteristic of a ‘new’ form ofhost community.

An improved research design

There are, in summary, three elements in Healyand Birrell’s research that require attention.First, we need to use ancestry data, rather thanbirthplace data, because we need to be able tocreate subsets of the Australian born populationaccording to their ethnic background. Secondly,we need to conduct the analyses at two levels ofaggregation: one needs to reflect the aggregationlevels used in international research into ethnicsegregation for comparative purposes and theother needs to distinguish between the earlierpost war immigrants and more recent immi-grants. Thirdly, we need to adopt methods ofanalysis that will produce measures of ethnicmix brought about in part through spatial assim-ilation. To achieve these three goals this paper:(1) reports on measurements of ethnic concen-tration using both birthplace and ancestry data;(2) aggregates the ethnicity data in eight differ-ent ways, three involving birthplace data, andfive involving ancestry data; and (3) adopts twoanalytical methodologies, the threshold andtypology approaches, that enable us to measureethnic mix and spatial assimilation. It is to thetechnical aspects of this approach that we nowturn.

The ancestry data

The Australian census authorities have onlycollected ancestry data twice, in 1986 and 2001.Because of flaws in the 1986 ancestry data (forFigure 1 Evolution of ethnic concentrations.

Is Sydney a Divided City Ethnically? 361

© Institute of Australian Geographers 2004

a detailed review, see Kunz, 2003), the 2001ancestry data take on special significance. Becauseof the problems with the 1986 data, relativelylittle use was made of that data set (though seeJones, 1991; Burnley, 1994) and as a conse-quence, ancestry was not included in the 1991Census (Kunz and Costello, 2003, 1). Testing ofa proposed ancestry question for the 1996 cen-sus again raised issues of data quality and itwas not used. Subsequent research and testing‘showed that acceptable identification of thesegroups could be achieved with a combination ofan ancestry question and a question on whethera person’s parents were born in Australia oroverseas’ (Kunz and Costello, 2003, 2). Conse-quently, a question about ancestry was includedin the 2001 Census. In contrast to the AustralianBureau of Statistics’ 1986 post-census evalu-ation report, the 2001 report sustains the qualityof the data.

The 2001 Australian census allows us toderive an ethnic classification of the populationfrom ancestry data, and the period of 30 yearssince the immigration policy was substantiallychanged in 1973 provides an opportunity tomake an initial assessment of the geographicaloutcomes of Australia’s experience of a multi-cultural approach to immigration. If the delayedassimilation model is correct, and operating inthe way we might expect, we should see the firstevidence of third generation immigrant Austra-lians living outside their ethnic enclaves, awayfrom their ethnic group’s areas of concentration,and mixing with the host community. This is notto say that they will be distributed proportion-ately in the same manner as the host society. Toachieve a spatial distribution which matches thatof the host community takes longer to achieveas the ethnic population gradually acquires (overone generation and more) the same socio-economic mix as the host community.

The 2001 census provides five subsets ofancestry data based on the place of parent’sbirthplace:

1. Both parents were born overseas.2. Female parent was born overseas.

3. Male parent was born overseas.4. Both parents were born within Australia.5. Place of parents’ birth was not stated.

These tables record the first two ancestryresponses reported and account for approxi-mately 94% of the total (i.e., 6% of the respond-ents gave more than two ancestries, hence for6%, additional ancestries were lost).

Following the approach adopted by the cen-sus (Kunz, 2003), where both parents are bornoverseas, the ancestry of the respondent is eitherfirst generation Australian (if the respondent wasborn overseas) or a second generation Australian(if the respondent was born in Australia). Wedo not know, from the CD Summary Files, thebirthplace of respondents in the ancestry tableso we refer to members of this sub-group as‘1st and 2nd (1st /2nd) generation’ Australians.Where both parents were born in Australia, therespondent’s ancestry is either third generationor some subsequent generation Australian.Because of lack of information about whichsubsequent generation they are, the members ofthis sub-group are referred to as ‘3rd Plus gener-ation’ Australians. Persons who have one parentborn overseas are considered as second gener-ation Australians, and are included as membersof the ‘1st and 2nd generation’ Australian group.Subtracting those born overseas (the first gener-ation Australians) from the ‘1st and 2nd gener-ation’ Australians adjusted from responses topopulation numbers provides a crude estimate ofthe total number of second generation Austra-lians. However, the same calculation cannot beundertaken for individual immigrant groupsbecause it is problematic for groups like theChinese. Many Australian residents of Chineseancestry were born outside China and hence arenot classified as Chinese in the birthplace data,though they are for ancestry purposes. Thosewho did not specify either their ancestry or theirparents’ birthplace have been excluded

2

. Thetotals of responses for ancestry in this study arethus the sum of (1) the first and second genera-tion Australians and (2) the third-plus generationAustralians. Similarly, for the birthplace data,

362 Australian Geographical Studies

© Institute of Australian Geographers 2004

which have been analysed for comparison withthe ancestry data, those who did not specifytheir birthplace were excluded from the totals.

The study area for this research was theSydney Statistical Division and the analyseswere conducted on the smallest census units inAustralia, the collection districts (CDs). All ofthe data referred to in the analyses and tablesare from the Australian Bureau of StatisticsCDATA2001. The decision to work at the levelof the CDs distinguishes this study from thoseof Burnley (1994) and Healy and Birrell (2003),who worked with Local Government Area (LGA)aggregated data. Research on ethnic enclavesmust be undertaken at spatial scales much smallerthan the LGAs, for two main reasons. First,many of the LGAs, especially in suburban Syd-ney, are large relative to the size of some of theethnic groups being studied. Second, data forLGAs can identify those parts of Sydney wheremembers of particular groups are concentratedin a general sense, but cannot address the issueof whether they live in neighbourhoods whereeither their own ethnic group, a range of ethnicgroups, or the host society are dominant. CDs,on the other hand, are well-suited to this task:they averaged 561 persons in Sydney in 2001.

The host population in this study is definedas all those of Australian, English, Welsh, Scot-tish, Irish, New Zealand, American, or Canadianbirth or ancestry

3

. The reason for defining thehost population in this manner was to identify asclosely as possible the same populations both interms of birthplace and ancestry. However theNESB group for birthplace does not containexactly the same population as the non-Anglo-Celtic population for ancestry, because many inthe latter group were born in Australia and aretherefore included in the host community for thebirthplace data.

Aggregation of ethnic data

The second technical issue, as we have arguedabove, is how we aggregated the data into ethnicgroups. The most common division of such datain the United Kingdom, Canada, New Zealand,Australia, and the United States of America is

between Whites, Blacks, Asians and Indigenousgroups. Certainly an immigrant /non-immigrantsplit is rare in ethnic segregation research,because immigration areas are seen as transitorylocations for most individual immigrants. Toovercome this problem, we have aggregated thebirthplace and ancestry data in eight differentways: three ways for the birthplace data and fivefor the ancestry data. Starting with a binarysplit, we have replicated Healy and Birrell’shost society and remainder aggregations. For thebirthplace data these are the ESB (English-Speaking Background) and NESB components,as defined above. With the ancestry data they arethe AC (Anglo-Celtic) and NAC (Non-Anglo-Celtic groups). The host society (ESB and AC)categories are not further subdivided. In thethird and fourth sets of analyses, the minoritycommunities are split first into those from Asianand white backgrounds

4

, then in the fifth andsixth analyses, those two groups in the ancestrydata are split by generations, and then, finally,according to geographical origins, using the 25(birthplace) and 27 (ancestry) separate non-hostcategories available at the CD level (the seventhand eighth analyses).

The expectation from these eight differentways of aggregating the ethnicity data is that themore disaggregated the data becomes, the lowerthe levels of ethnic concentration will be. Forexample, if all of the people from the individualAsian nations in a census area are added together,they will form a higher proportion of the totalpopulation in that area than will any singlegroup by itself. Hence, if bifurcation did exist inSydney, it would be more likely measured bytotalling all the immigrants, rather than at lowerlevels of aggregation. However, it is possiblethat the ethnic concentration for a highly aggre-gated group is primarily due to a subset of thatgroup. Hence we need to investigate a mix ofdifferent aggregations.

Analyses

The third and final technical issue is that of themethods of analysis. Both the index and factor-ial ecology approaches previously used for the

Is Sydney a Divided City Ethnically? 363

© Institute of Australian Geographers 2004

analysis of ethnic group segregation have comein for criticism in recent years because of theirfocus on relativities. They identify levels andareas of ethnic concentration compared with onlygroup averages for the city or area being stud-ied. They say little, if anything, about absolutelevels of segregation and are, therefore, unsuitedto the assessment of change over time (becausethe averages themselves also change) or fromone place to another (where, again, the base —the averages — are different from one city toanother). To address this problem, absolutemeasures are required. Building on work byPhilpott (1978), Peach (1996) and Jargowsky(1997) (see also Johnston and Poulsen, 2000),Poulsen

et al

. (2002a) have developed the use ofsuch absolute measures. These are incorporatedinto two sets of analysis employed in this study.

The first is threshold analysis (Johnston andPoulsen, 2000; Johnston

et al

., 2003a), whichexamines the levels of concentration of eachgroup within a census area. The second is atypology (Poulsen

et al

., 2001b), which classi-fies areas (CDs in this case) according to theirethnic/host society composition. These allow amuch more rigorous evaluation of hypothesesregarding absolute levels of segregation — whichare the basis of Healy and Birrell’s claims.

Threshold analyses

Each ethnic group and the host society has itspercentage share of the total population withineach census sub-area identified (commonly at10, 20, 30, 40, . . . 100% threshold levels; theseintervals are necessarily arbitrary, but as long asa wide range is selected, the number of thresh-olds merely influences the detailed shapes ofprofiles; more threshold points may be justifiedin some cases). The percentage of a relevantgroup’s population living in districts at eachthreshold interval value across the entire city isthen calculated, and a graph of those levels ofconcentration prepared. For example, Figure 2shows that 30% of the type 3 population lives inCDs where they comprise 10% or more of theCD population, and 10% live in CDs where theyare 70% or more of the local population.

Five ideal-typical concentration curves areidentified in Figure 2 to illustrate different levelsof segregation. If Sydney were a ‘divided city’the threshold analyses should display a type 1curve for the minority groups (Figure 2). This isa convex curve that must be to the right of thatminority group’s share of the city’s total popu-lation, showing greater concentration in areas(CDs) with relatively high percentages of thatgroup in their population than would occur witha random distribution. This form of curve is typ-ical of African Americans in major UnitedStates cities and indicates high levels of segre-gation (as shown in Poulsen

et al

., 2002b). Type 2 curves, on the other hand, would be

the norm where a group’s members are distrib-uted equally across all levels of concentration:equal proportions live in highly segregated andnon-segregated areas. This curve often takes ona slightly convex form as displayed for Hispanicsin American cities and Maori in New Zealandcities (Johnston

et al

., 2003b; 2003c). A further form of segregation is indicated by

a type 3 curve where only a small proportion ofthe group live in highly segregated areas. Researchin America and New Zealand has associated thisform of curve with Asian populations.

A non-segregated structure is indicated by atype 4 curve, where the population is distributed

Figure 2 Examples of different types of concentration curves.

364 Australian Geographical Studies

© Institute of Australian Geographers 2004

equally across the areas according to its share ofthe city’s total population

5

. The type 5 curve in Figure 2 is an example of

a stepped variant of a type 2 curve. Here, thelevels of concentration do not form a contin-uum; such stepping may be observed when thenumber of observation units is relatively small(as in Johnston

et al

., 2004) or when there is anested set of concentrations.

Residential area typologies

The typology analyses set up a generic systemfor classifying areas according to their ethnicgroup(s)/host society mix (see Poulsen

et al

.,2001b; 2002a). It is an absolute system of clas-sification, involving several steps. The first stepdivides the collection districts into those wherethe host society is dominant (i.e., comprisingmore than 50% of the population), and thosewhere it forms a minority

6

. The second stepinvolves further subdivision on the basis of moredetailed information about their ethnic makeup,as follows.

Areas where the host society is in the major-ity (termed host communities) are subdividedinto two types (Figure 3):

1.

Isolated host communities

where the hostsociety forms 80% or more of the total pop-ulation, and is substantially isolated fromthe ethnic groups that occupy other parts ofthe city. Following Marcuse and van Kempen(2000), these areas are also termed ‘whitecitadels’.

2.

Non-isolated host communities

where thehost society forms 50–80% of the local pop-ulation. Members of the host society in suchareas live among a majority drawn from theirco-ethnics, but are not as isolated from mem-bers of (at least some) other ethnic groups asis the case with those living in the first type.

Areas where members of the host society arein a minority are divided into four types accord-ing to the percentage of their populations whoare members of the host society and the per-centage of the various ethnic minority groupspresent:

1.

Associated assimilation-pluralism enclaves

are the areas with the greatest ethnic mix.Members of the host society are in a sub-stantial minority (30–50% of the total) andminorities form between 50% and 70%.

2.

Mixed-minority enclaves

where the minority(i.e., non-host) ethnic groups form a substan-tial majority of the total but no one minoritygroup is more than twice the size of all otherminorities. They are areas of relatively highmixing of minority groups with some hostsociety presence.

3.

Polarised enclaves

are areas where the dom-inant minority group has more than twicethe number of all other minorities combined.The minorities, as in group 4, exceed 70% ofthe population. At a minimum the dominantminority forms 66.6% of the minority popu-lation and up to a maximum of 100%.

4.

Ghettos (extreme polarised enclaves)

notonly have to be polarised enclaves, but, inaddition, the dominant group must (1) exceed60% of the total population in the relevantCDs and (2) have 30% or more of its totalpopulation in the city living in these enclaves.For the African American in the UnitedStates these are the ghettos, for the Hispanicsthey are the barrios (Johnston

et al

., 2003b).

Figure 3 Typology analysis types.

Is Sydney a Divided City Ethnically? 365

© Institute of Australian Geographers 2004

There are three types of area where ethnicgroups and the host population live in mixedmilieux: mixed enclaves; associated assimilatedenclaves; and non-isolated host communities(Figure 3). Where a city is characterised by mix,the majority of the ethnic population and a highproportion of the host population will livewithin these three types of areas. But, if a cityis ethnically polarised, the majority of at leastone of the ethnic groups will live in one or otherof the two types of polarised enclaves, and ahigh proportion of the host population will livein isolated host communities (the white citadels).

Sydney — a divided city?

The two methodologies described above arewell suited to critically evaluate Healy andBirrell’s contention that Sydney is becoming anethnically divided city. Using both birthplaceand ancestry data, we first assess whether theircontention is a function of their use of birthplacedata alone. (Only ancestry data, which allow thesettlement patterns of second and later gener-ation immigrants to be mapped, can address thequestion of continuing segregation or desegre-gation of those groups into the second andsubsequent generations.) Second, only by disag-gregating the ancestry data into regional andnational groups can we evaluate ethnic segrega-tion at a scale which is in keeping with researchconducted internationally (Healy and Birrellworked at the most aggregated levels of ethnicity— the NESB and Australian-born). Third, theirfocus on segregation as opposed to mix fails toaddress the issue of whether ethnic enclaves

are transitory (Jones, 1996) or not (Healy andBirrell, 2003).

Whether birthplace or ancestry data aredeployed, Sydney’s population is dominated byits host population (Table 1), although this issignificantly smaller according to the ancestrydata compared with the birthplace data

7

. How-ever, immigration from NESB/NAC sources overthe last 50 years has considerably altered Sydney’smix from a relatively homogeneous society withan estimated 96% of AC stock at the 1947 census(Jones, 1996) to 75% Australian born or 64%AC by ancestry in 2001. The birthplace dataindicate that approximately 25% of Sydney’spopulation were NESB immigrants. As a crudeestimate of the number of second generationAustralians, we can subtract the NESB group(first generation by birthplace) from the ‘1stand 2nd generation’ group based on ancestryresponses

8

. This indicates for the NAC popu-lation that 61% are first generation, 30% are sec-ond, and 10% third plus generation Australians.That the offspring of immigrants only accountfor half of the number of immigrants (NESBgroup) is seen as a combination of recency ofarrival, a bias towards family immigration, and theabandonment by immigrants, or their descend-ants, of their original ancestry in favour ofAustralian ancestry. An average of 561 personslive in each census collection district. The averagenumber of ancestry responses is 651. Given thateach person had their first two listed ancestriescoded, a ratio of 1.16 ancestry responses perperson indicates that most of Sydney’s popu-lation associate themselves with a single ancestry.

Table 1 Differences between birthplace and ancestry data.

Birthplace Ancestry Birthplace (%) Ancestry (%)

Host 2 756 648 2 690 064 74.7 63.6NESB 931 371 25.3NAC 1st and 2nd Generation 1 385 168 32.8NAC 3rd Plus Generation 151 314 3.6Total 3 688 019 4 226 546Per CD 561 651

366 Australian Geographical Studies

© Institute of Australian Geographers 2004

An initial indicator of the degree of spatialsegregation for any group is the proportion of itspopulation living in areas where they form themajority. By birthplace, the host community(the Australian born and ESB immigrants) formthe majority population in 90% of the collectiondistricts. This figure drops to 73% for the ACgroup based on the ancestry data (Table 2)

9

. Theproportion of the host /AC group that live incollection districts where they are the majority(95% and 86%) is also extremely high. This isnot the case for the NESB (birthplace data) orthe NAC populations (ancestry data). Only 24%of NESB immigrants, as defined by the birth-place data, live in areas where they are in themajority, rising to 44% of the NAC (first andsecond generation) group based on ancestry. Thisdifference is because the host community bybirthplace contains the descendants of NESBimmigrants, whereas they are part of the NACgroup in the ancestry analysis. By marked con-trast, only 0.1% of the non-host ‘3rd plus gener-ation’ live in areas where they are in the majority.

Threshold analyses

Threshold analyses identify the general patternof segregation for each group, which can beset against the ideal-typical patterns shown inFigure 2. Eight different sets of threshold analy-ses were run. In the first, the birthplace data

were split into members of the host communityand NESB immigrants. For the second analysisthe ancestry data equivalent subsets were thosewith Anglo-Celtic (AC) and non Anglo-Celtic(NAC) backgrounds. Figure 4 shows a type 2curve for the NESB group, indicating a fairlyeven population distribution across all thresholdlevels. Some 60% of those born outside Aus-tralia live in areas where such people form atleast 30% of the total, for example, and just over20% live in areas where they form a majority.However, very few live in areas where NESB

Table 2 Maximum levels of concentration and majority status.

Maximum in a census collection district(%)

Collection districts where group is the majority population(%)

Groups living in collection districts wherethey are the majority population(%)

BirthplaceHost 100 89.6 94.6NESB 87.4 9.6 23.7

AncestryHost (AC) 97.2 72.6 85.6NAC 1st and 2nd generation 97.0 23.1 44.0NAC 3rd generation 94.9 0.0 0.1

Figure 4 Birthplace — levels of concentration of the hostand NESB groups (threshold analysis 1).

Is Sydney a Divided City Ethnically? 367

© Institute of Australian Geographers 2004

people predominate. Certainly this is not thetype 1 curve, associated with groups living inhighly segregated areas. By contrast, the hostpopulation is much more segregated, with some70% living in collection districts where at least70% of the population is drawn from the hostcomponent. But, because 75% of Sydney’s popu-lation were members of the host community in2001, this pattern is entirely consistent with arandom allocation of its members to collectiondistricts, with relatively few of them living inareas with very high percentages of host societymembers.

The pattern shown in Figure 4 is largelyrepeated in the analyses of the ancestry data(Figure 5), but with members of the AC hostsociety slightly more segregated than a randomallocation would produce. They form 62% ofthe total, but over 70% live in areas where theAC host society forms at least 60% of the total.There is very little extreme segregation, how-ever, and virtually none lived in areas that were90% AC host society in their composition.Those in the NAC group have a type 2 curve:some live in areas of considerable segregation— 30% in areas that are at least 60% NAC, forexample, but many are in areas where NAC areonly a small proportion of the total.

In the third set of analyses, the birthplacedata, the NESB group are divided into thewhite-NESB and the Asian populations. Forthe fourth set of analyses, the ancestry data, theNAC group are divided into the white-NAC andAsian populations. Figure 6, for the birthplacedata, shows that neither NESB group displaysany extreme levels of concentration in Sydney:among the white-NESB group, virtually nonelive in areas where this group forms more than20% of the total. Although there is slightly moresegregation of the Asian group at the lowerthresholds (given that they form only 13% ofSydney’s total population), only 5% live in areaswhere Asians form half or more of the totalpopulation. The observed curves are of type 3,which suggests that the groups are, in general,distributed according to their share of the totalpopulation across most areas with only smallproportions in areas of high concentration. Thesame type of curve and conclusions are revealedby the ancestry data (Figure 7). However, thereare two differences between the two data sets.First, the levels of concentration for both non-host group profiles (white-NAC and Asian) aremuch higher for the ancestry data which partlyreflects the fact that these groups are larger thantheir birthplace counterparts (i.e., there are more

Figure 5 Ancestry — levels of concentration of the hostand NAC groups (threshold analysis 2).

Figure 6 Birthplace — splitting the NESB group (thresh-old analysis 3).

368 Australian Geographical Studies

© Institute of Australian Geographers 2004

people with Asian ancestry than with Asianbirthplaces). Second, there is no major differ-ence in the profiles between the white-NAC andAsian populations — the two virtually overlap— whereas the two curves for the birthplacedata indicate slightly more segregation amongAsians than among non-Asians.

The fifth set of threshold analyses is basedentirely on the ancestry data with the NACgroup subdivided by generations. This subsetresults in a graph that is very similar to that forthe NAC group overall. Figure 8 displays a type2 curve for the NAC ‘1st/2nd generation’ group,indicating that some of these relatively recentmigrants and their offspring live in areas wherethey form a large component of the population:40% of them live in areas where they are in amajority, for example, and nearly one-fifthreside in areas where they form 70% of the total.On the other hand, there is virtually no segrega-tion of the relatively small number (4% of thetotal population) of ‘3rd plus generation’ Aus-tralians of NAC ancestry (none of them live inareas where they comprise even 10% of the totalpopulation). Likewise, in the sixth set of analy-ses, there is no major difference in the profilesbetween the white-NAC and Asian populationswhen they are subdivided on the basis of gener-

ations (Figure 9). However, there is the expecteddifference between the profiles for the ‘1st/2ndand 3rd plus generation’ groups in Figures 8and 9.

In the final sets of threshold analyses, thegeographically-defined birthplace and ancestrygroups have been used. Potentially, the white-NESB or white-NAC and Asian populationscould contain a national grouping which displayeda type 1 curve, in that they could potentially be

Figure 7 Ancestry — splitting the NAC group (thresholdanalysis 4).

Figure 8 Ancestry — the NAC group divided by genera-tions (threshold analysis 5).

Figure 9 Ancestry — the NAC group split by ethnicity andgenerations (threshold analysis 6).

Is Sydney a Divided City Ethnically? 369

© Institute of Australian Geographers 2004

clustered into a small number of collectiondistricts. Figure 10 displays the seventh set ofanalyses and contrasts, with the host communityprofile on the right, and all 25 other nationalgroups on the left. No ethnic group, by birth-place, displays a highly segregated type 1 curve:all of these curves are types 3 or 4. The sameresults are also conveyed in Figure 11, the eighthset of analyses, based on 27 national groups forthe ancestry data. Again there is little evidenceof strong segregation for any one group; for the

great majority, less than 20% live in areas wheretheir co-ethnics form even 20% of the popula-tion total and in only two groups do more than10% live in areas shared with at least 30% ofpeople with the same origins. There is little, ifany, evidence here of a city divided betweenethnic and host society fragments as suggestedby Healy and Birrell (2003).

Typology analyses

In the final set of analyses, the data are dividedinto the same birthplace and ancestry groups asfor the preceding eight threshold analyses

10

. Theresults bear out the findings of the previous sec-tion. At the most aggregated level, comparablewith that of Healy and Birrell’s (2003) analysis,Tables 3 and 4 present the results for the birth-place and ancestry data, the first and secondanalyses, where the NESB and NAC popula-tions are compared with the host society. Table 3shows, for birthplace, that most (95%) of thehost community live in areas where members ofthat community are in the majority. They arealmost evenly split between those who live inthe white citadels and in mixed non-isolatedhost communities. For the NESB population,76% live in areas where the host community arein the majority. Of the 24% that live in minorityareas, only 3% live in polarised enclaves andnone in the most extreme form of segregation,

Figure 10 The national birthplace groups (thresholdanalysis 7).

Figure 11 The national ancestry groups (threshold analysis 8).

Table 3 Typology analyses at the maximum aggregationlevel (Birthplace data, analysis 1).

Birthplace Data Host(%)

NESB(%)

Totals(%)

Extreme Polarised Enclave (ghetto)

0.0 0.0 0.0

Polarised Enclave 0.3 3.2 1.0Mixed Enclave 0.2 1.1 0.5Associated Assimilation Pluralism Enclave

4.8 19.1 8.4

Non-isolated Host Community

39.7 56.7 43.7

Isolated Host Community (white citadel)

55.1 19.8 46.3

370 Australian Geographical Studies

© Institute of Australian Geographers 2004

the ghetto. Clearly these results do not supportthe bifurcation claim that ‘there are now twoSydney’s — one increasingly dominated by lowto moderate-income non-English-speaking migrantcommunities in the West and South-West andthe other comprised of established inner afflu-ent areas and predominantly English-speaking‘aspirational’ areas in the metropolitan periph-ery’ (Healy and Birrell, 2003, 65).

The ancestry data (Table 4) indicate a degreeof mixing of the host (AC) society with NACgroups which is not present in the birthplacedata: fewer of the host community live in thewhite citadels, and more in non-isolated hostcommunities. Nevertheless, 7% of the AC popu-lation and 31% of the NAC population do live

in areas of extreme polarisation. In other words,use of the ancestry data does identify someextreme segregation, affecting almost one-thirdof the NAC population. For the NAC popula-tion, therefore, the distribution is very differentfrom that measured for the NESB birthplacepopulation. But this level of segregation is notthe primary pattern for the NAC group. Theessential point is that 60% of the NAC grouplive in the three types of mixed areas. In addi-tion another 9% live in white citadels. The factthat 69% of the NAC group live in some formof mix with the host community refutes anyclaim of bifurcation.

Conducting segregation studies at the NESBor NAC levels of aggregation, as undertakenabove, however, is somewhat misleading.Tables 5 and 6 show results for the third andfourth analyses. Here the NESB and NAC groupsare split, respectively, into their two main com-ponent populations, white-NESB and Asian.Table 5, using birthplace data, shows that 85%of the white-NESB population live in areaswhere the host community are in the majority,i.e., they are not segregated from the hostcommunity to any substantial degree. Withinthe host community areas more than twice thenumber live in relatively mixed, non-isolatedareas than in isolated host communities (60% asagainst 25%). For the Asians, 69% live in hostcommunities. Again, none of the white-NESBnor Asian population live in ghettos, with 15%of white-NESB and 31% of Asians living inminority areas; only 0.6% and 3.9% respectivelywere living in polarised enclaves.

Table 4 Typology analyses at the maximum aggregationlevel (Ancestry data, analysis 2).

Ancestry Data AC (%)

NAC (%)

Totals (%)

Extreme Polarised Enclave (ghetto)

7.2 31.5 16.2

Polarised Enclave 0.0 0.0 0.0Mixed Enclave 0.0 0.0 0.0Associated Assimilation Pluralism Enclave

6.4 13.2 8.9

Non-isolated Host Community

56.2 46.5 52.6

Isolated Host Community (white citadel)

30.2 8.9 22.3

Table 5 Typology analyses with the NESB and NAC populations split between whites and Asians (Birthplace data, analysis 3).

Birthplace Data Host(%)

White NESB(%)

Asian(%)

Totals(%)

Extreme Polarised Enclave (ghetto) 0.0 0.0 0.0 0.0Polarised Enclave 0.2 0.6 3.9 0.7Mixed Enclave 0.3 1.7 2.6 0.9Associated Assimilation Pluralism Enclave 4.7 12.8 24.6 8.3Non-isolated Host Community 39.7 59.7 54.0 43.7Isolated Host Community (white citadel) 55.1 25.2 14.9 46.3

Is Sydney a Divided City Ethnically? 371

© Institute of Australian Geographers 2004

Using ancestry data (Table 6), however, thestructures observed for the NAC data at thehighest level of aggregation in the previous setof analyses on ancestry (Table 3) are not sus-tained. Most importantly, neither the white-NACnor the Asian populations, taken separately,form ghettos in Sydney. Even at the level ofpolarised enclaves only 7% of the white-NACpopulation and 13% of the Asian population livein those types of areas. Some 40% of the white-NAC population and 55% of the Asians live inminority areas. But most importantly, disaggre-gating the combined ethnic groups to the levelof even major ethnic groupings transforms theapparently high levels of segregation in Sydneyto high levels of mix. The primary feature ofSydney’s ethnic geography is that 80% of thewhite-NAC population live in the three mixedtypes of concentrations, as do 83% of Asians.There are very few relatively exclusive residen-

tial areas where members of one ethnic groupdominate — even when using the coarse-grained division of white-NAC and Asian.Where ethnic concentrations do occur, theycomprise mixtures of both of these types.

In the fifth set of analyses, the NAC popu-lation are disaggregated into a ‘1st /2nd genera-tion’ group and a ‘3rd plus generation’ group. Thisanalysis represents another way to make subsetsof the NAC population (Tables 3 and 4). Table 7demonstrates once more that breaking the NACpopulation down into subgroups highlights veryhigh levels of ethnic mix within small (CD) areas.Of the NAC ‘1st /2nd generation’ population,67% live in mixed areas, as do 70% of the NAC‘3rd plus generation’. Only the NAC ‘1st/2ndgeneration’ population displays a high level ofconcentration in polarised enclaves (26%). As withthe threshold analyses, these typologies provideample evidence that the ‘3rd plus generation’ are

Table 6 Typology analyses with the NESB and NAC populations split between whites and Asians (Ancestry data, analysis 4).

Ancestry Data AC(%)

White NAC(%)

Asian(%)

Totals(%)

Extreme Polarised Enclave (ghetto) 0.0 0.0 0.0 0.0Polarised Enclave 1.3 6.8 12.7 3.8Mixed Enclave 1.5 8.4 14.4 4.5Associated Assimilation Pluralism Enclave 9.4 24.8 27.7 14.6Non-isolated Host Community 56.2 46.8 41.4 52.6Isolated Host Community (white citadel) 30.2 13.1 3.8 22.3

Table 7 Typology analysis with the NAC population split on the basis of generations (Ancestry data, analysis 5)*.

AC(%)

NAC12g(%)

NAC3pg(%)

Totals(%)

Extreme Polarised Enclave (ghetto) 0.0 0.0 0.0 0.0Polarised Enclave 4.4 25.5 7.1 11.6Mixed Enclave 0.0 0.0 0.1 0.0Associated Assimilation Pluralism Enclave 9.2 21.8 13.8 13.6Non-isolated Host Community 56.2 45.3 56.4 52.6Isolated Host Community (white citadel) 30.2 7.3 22.8 22.3

* AC = Anglo Celtic; NAC12g = Non-Anglo Celtic 1st /2nd generation; NAC3pg = Non-Anglo Celtic 3rd plus generation.

372 Australian Geographical Studies

© Institute of Australian Geographers 2004

moving out of immigrant enclaves through theprocess of spatial assimilation. A further split-ting of this generational data into white-NACand Asian populations in the sixth set of analy-ses (Table 8) indicates mix is the primary pat-tern, and levels of concentration in polarisedenclaves are halved. In other words, as onedisaggregates the data, the dominant pattern ofethnic mix increases. For those Asians in the‘1st/2nd generation’ group, the percentage dropsto 11%, whereas, for the white-NACs it is onlyhalf that percentage. For the ‘3rd plus gener-ation’ the numbers are even smaller. This furtherreinforces the finding that immigrants’ grand-children are displaying high levels of spatialassimilation.

Finally, the typology analyses were undertakenon the geographical birthplace and ancestrydata to make certain that no individual nationalgroup, on its own, displayed a high level ofsegregation. Potentially, an individual nationalgroup could, if distributed across a limited setof collection districts, form a highly polarisedenclave. Table 9 indicates that none of the 25NAC ethnic groups defined by birthplace in theseventh set of analyses lives in ghettos. Simi-larly, a negligible percentage live in polarisedenclaves. The sharing of residential locationsdominates the structure, with no national grouphaving less than 50% of its population living

in areas where the host community is in themajority. With the exception of the Dutch andGermans, most of these immigrant groups livein host communities that are relatively mixed(non-isolated host communities) than in themore exclusive white citadels. Of those livingin minority areas, a higher percentage live inmixed enclaves, except for immigrants from thePhilippines, Vietnam, Malta and Yugoslavia. Ingeneral, the Asian populations are more likelyto have a larger percentage of their populationsliving in those mixed enclaves.

By ancestry, the eighth and final set of analy-ses (Table 10), illustrate the same conclusions.The dominance of mix prevails. For all but nineof the 27 ancestral groups more than half theirpopulations live in communities where the hostAC population is in the majority. Only those ofVietnamese (18%), Lebanese (33%), Macedo-nian (34%) and Turkish (35%) ancestry departsignificantly from this pattern. Moreover, nonational ethnic ancestry group forms a ghettoand there are very few people living in polarisedenclaves. At the other extreme, the percentageliving in white citadels is also very small, withthe largest percentages being less than 33%.Mix, clearly, is the dominant feature of mostnational groups in Sydney, with 80% or moreliving in one of the three mixed types of concen-trations where they share those areas with other

Table 8 Typology analysis with the NAC population split on the basis of ethnic group and generations (Ancestry data,analysis 6)*.

AC(%)

WhiteNAC12g(%)

Asian12g(%)

WhiteNAC3pg(%)

Asian3pg(%)

Totals(%)

Extreme Polarised Enclave (ghetto) 1.3 0.2 3.6 1.9 1.9 2.1Polarised Enclave 1.1 6.5 11.0 1.9 2.8 3.3Mixed Enclave 1.7 10.5 14.8 2.8 4.4 5.1Associated Assimilation Pluralism Enclave 9.5 27.0 25.6 14.1 13.9 14.7Non-isolated Host Community 56.2 45.0 41.7 56.4 56.3 52.6Isolated Host Community (white citadel) 30.2 11.2 3.3 22.9 20.8 22.3

* AC = Anglo Celtic; NAC12g = Non-Anglo Celtic 1st/2nd generation; NAC3pg = Non-Anglo Celtic 3rd plus generation.

Is Sydney a Divided City Ethnically? 373

© Institute of Australian Geographers 2004

ethnic groups, or with the host communities.Top of the list for the mixed enclaves by ances-try are the Vietnamese (75%), followed by theTurkish (57%) and Macedonian (56%).

Conclusions

Healy and Birrell (2003) are correct in identify-ing the continued expansion of areas of immigrantconcentration. Such a pattern of development isconsistent with immigration theory regardingthe early period of their acceptance into the newsociety. All that is different over the past threedecades, compared with the two previous dec-

ades under the former White Australia policy, isthat more recent immigrant streams derive froma very much wider range of source areas. Theyare more visible than the previous generations ofESB/AC immigrants. But we cannot make thejump in logic from the expansion of numbers ofNESB persons living in poorer local governmentareas to the notion of bifurcation occurringbetween the host and the immigrant communi-ties. Nor can we make the next jump in logicand claim that the current ethnic concentrationsare likely to be a permanent feature, because thedisaggregated data of the NESB community

Table 9 Typology analysis based on national birthplace data (analysis 7).

Extreme Polarised Enclaves (Ghetto)(%)

Polarised Enclave(%)

Mixed Enclave(%)

Associated Assimilation Pluralism Enclave(%)

Non-isolated Host Community(%)

Isolated Host Community (white Citadels)(%)

Host (Anglo Celtic birthplace) 0.0 0.0 2.7 2.5 39.7 55.1Fiji 0.0 0.3 12.0 10.6 59.9 17.1China&HK 0.0 0.1 18.5 13.0 55.0 13.5India 0.0 0.0 16.3 8.1 54.0 21.6Indonesia 0.0 0.0 23.0 10.6 51.6 14.8Korea (Rep) 0.0 0.0 23.2 9.2 53.9 13.7Malaysia 0.0 0.0 9.0 5.7 56.1 29.2Philippines 0.0 0.3 10.4 10.7 59.9 18.8Singapore 0.0 0.0 10.6 4.7 54.1 30.6Sri Lanka 0.0 0.1 15.6 9.6 56.7 17.9Vietnam 0.0 0.2 20.8 27.6 47.0 4.3Croatia 0.0 0.4 10.3 9.8 54.9 24.6France 0.0 0.1 3.5 3.3 48.0 45.1Germany 0.0 0.1 2.9 3.4 43.0 50.5Greece 0.0 0.1 9.1 4.0 68.6 18.2Italy 0.0 0.1 4.4 5.6 64.8 25.2Macedonia 0.0 0.1 13.0 8.5 65.0 13.3Malta 0.0 0.0 2.4 3.5 61.7 32.4Netherlands 0.0 0.0 1.8 1.8 33.7 62.7Poland 0.0 0.2 7.2 6.4 58.3 27.9Yugoslavia 0.0 1.0 14.0 14.4 49.6 21.0Egypt 0.0 0.1 7.1 6.3 60.1 26.4Lebanon 0.0 0.1 15.0 6.0 69.3 9.6Turkey 0.0 0.4 23.5 10.0 54.1 11.9South Africa 0.0 0.0 2.2 2.1 53.9 41.8Elsewhere 0.0 1.0 11.2 12.7 52.8 22.3Totals 0.0 0.1 5.2 4.6 43.7 46.3

374 Australian Geographical Studies

© Institute of Australian Geographers 2004

even at the most general level indicates other-wise. Healy and Birrell’s research design doesnot provide the evidence to substantiate eitherclaim. Indeed, on the evidence presented here, itcannot.

In general, the ancestry data suggest thatthere are approximately twice as many people ofNAC ethnic backgrounds living in ethnic con-centrations than that shown by the birthplace

data. The threshold analyses have failed to indi-cate a type 1 curve, the high segregation curve,for any ethnic group at any level of aggregation.Our findings, from the typology analyses at thehost and NESB level of aggregation, show thatthe primary feature of the NESB population isthat they live in areas of mix, not in polarisedenclaves. A slightly modified conclusion isreached when using ancestry data at the AC and

Table 10 Typology analysis based on national ancestry data (analysis 8).

Extreme Polarised Enclaves (Ghetto)(%)

PolarisedEnclave(%)

Mixed Enclave(%)

Associated Assimilation PluralismEnclave(%)

Non-isolated Host Community(%)

Isolated Host Community (white Citadels)(%)

Host (Anglo Celtic ancestry) 0.0 0.0 10.5 3.1 56.2 30.2Maori 0.0 0.1 28.2 6.7 51.1 13.9Other Oceania 0.0 0.1 40.0 7.0 47.2 5.7Other Oceania Australians 0.0 2.7 22.3 4.2 52.7 18.1Chinese 0.0 1.0 42.5 11.8 41.6 3.1Filipino 0.0 0.1 37.3 10.7 46.5 5.3Indian 0.0 0.2 38.7 6.3 50.4 4.4Indonesian 0.0 0.9 41.3 9.1 43.2 5.5North East Asian Other 0.0 0.4 34.8 9.3 51.6 3.9South East Asian Other 0.0 0.2 53.5 5.7 36.2 4.4Sub Continent Asia Other 0.0 0.1 40.5 6.4 49.0 4.0Vietnamese 0.0 0.0 75.0 6.7 17.3 0.9Croatian 0.0 0.1 37.7 6.6 47.9 7.7Dutch 0.0 0.0 8.8 2.4 57.5 31.3German 0.0 0.0 12.6 3.4 58.3 25.7Greek 0.0 0.1 35.6 12.1 45.6 6.7Italian 0.0 0.0 28.3 8.7 51.9 11.0Macedonian 0.0 0.2 55.8 10.2 30.6 3.3Maltese 0.0 0.0 22.9 5.4 58.1 13.6North East and West Europe Other

0.0 0.1 11.9 3.2 60.4 24.4

Polish 0.0 0.1 24.4 4.5 57.7 13.3Russian 0.0 0.1 29.1 5.3 55.4 10.1Serbian 0.0 0.1 49.1 5.0 38.1 7.8South and East Europe Other 0.0 0.1 30.9 5.6 52.6 10.8Lebanese 0.0 0.3 51.6 15.4 30.7 2.1North Africa and Middle East other

0.0 0.1 52.2 6.8 37.7 3.2

Turkish 0.0 0.1 57.2 8.0 32.3 2.4Sub Saharan Africa 0.0 0.1 23.3 3.8 61.2 11.6Totals 0.0 0.1 20.1 5.0 52.6 22.3

Is Sydney a Divided City Ethnically? 375

© Institute of Australian Geographers 2004

NAC level. The NAC group primarily (68.5%)live in mixed areas, although there is a second-ary pattern in that 31.5% live in areas designatedas places of extreme polarisation. However,research into ethnic concentrations at this levelof aggregation is questionable in that it is notat the standard regional or national levels thatunderlie most segregation research. The NACgroup is made up of a wide range of differentcommunity groups, some of which are theimmigrants or descendants of the early postwar period, a group that all researchers agree,are undergoing spatial assimilation. Theiraggregation into a single large ‘ethnic’ group isinappropriate.

In breaking the birthplace or ancestry datadown from NESB or NAC populations to thewhite-NESB or white-NAC on the one hand,and the Asian on the other, the polarised struc-tures disappear. Mixing of these ethnic groups,among themselves and with the host society, isin every case the primary pattern of ethnicconcentration in Sydney. Significantly too, theanalyses presented here counter the argumentthat, in general, more recently arrived Asianpopulations are following a pattern of spatialand, by implication, social assimilation differ-ent to that followed by earlier post war NACgroups. Disaggregating the NAC population onthe basis of generations refutes any idea thatthe ‘3rd plus generations’ are entrapped in ethnicenclaves. Equally important, there is no differ-ence in behaviour between the ‘3rd plus gen-eration’ white-NAC and Asian populations. Thissame decline in segregation levels also occurswhen the population is further disaggregated onthe bases of national birthplace or nationalancestry types.

The primary feature of Sydney’s ethnic popu-lation to emerge from this study is one of resi-dential mixing,

not

of segregation or bifurcation.This finding reaffirms previous internationalcomparative research by Poulsen

et al

. (2002a).This mixture involves both the host communityand the minority communities and occurs amongNESB and NAC groups. Sydney is a leadingexemplar of ethnic residential mixing. The study

of how that mix has been achieved and of itsdynamics should form a key area of futureresearch. Australia’s multicultural immigrationpolicy, when evaluated against the expectationof delayed assimilation, appears to be working.The shift over the past 30 years by which areasthat were previously white citadels occupiedalmost exclusively by the AC population havebecome more ethnically mixed, means thatarguments regarding segregation and bifurcationbecome increasingly difficult to square with theevidence. Sydney appears to be moving towardsbeing a city in which most of the populationwill live in areas that are classified as eithernon-isolated host communities, associatedassimilation-pluralism communities, or mixedenclaves — a hybrid city (to use a term popularwith post-modern theorists) in which theongoing processes of hybridisation are clearlyreflected in the city’s geography.

Correspondence

: Dr. Mike Poulsen, Department of HumanGeography, Macquarie University, Sydney, New SouthWales, 2109, Australia. E-Mail: [email protected]

NOTES

1. One is the migration programme (100 000 to 110 000places in 2003/4) made up of a skilled migrationstream, a family migration stream and a special eligi-bility migrant stream. The other is the separatehumanitarian programme (12 000 places in 2003/4).In the skilled migration stream immigrants have tomeet a points score, which emphasises high-levelemployment skills, youth, and competency in English.The family migration stream (43 200 places in 2003/4or 41.1% of the migration program

87% werespouses in 2001/2) is selected on the basis of theirfamily relationship with their sponsor in Australia.

2. It is assumed that those who did not respond to theancestry question are comparable with those who didso within each census district. For all but a few censusdistricts this assumption is not an issue. These are areaswhere there are hotels, hospitals, nursing homes, andprisons, where management completed census forms.

3. This definition of the host population is an importantimprovement on earlier definitions where only Australian-born were included. As such, it brings the Australianresearch more into line with the American, Canadian,New Zealand and United Kingdom studies where thehost population is defined as the ‘white’ population.

376 Australian Geographical Studies

© Institute of Australian Geographers 2004

4. The host society cannot be subdivided in this model-ling approach because they are the comparative group.

5. Care must be taken in distinguishing between a type 1and a type 4 curve when a group’s share of the city’stotal population is very large (such as for the hostpopulation in Sydney). This is because both willappear on the right side of the graph. A type 1 curve,however, always has a marked convex form, whereas atype 4 curve takes a straight or concave form.

6. The key requirement here is that the host populationbe defined in a similar manner to that in comparable stud-ies (see note 3).

7. For the birthplace analyses between the host and theNESB groups, those born elsewhere were included inthe NESB group. In all other aggregations they form aseparate group. Similarly, for the ancestry data in theanalyses between the AC and the NAC groups, theIndigenous Australian, Maori, Other Australian Oce-ania, and Other Oceania responses were part of theNAC group. In all other aggregations they form separategroups. As previously noted, Asian, in this study,includes the ABS regions of northeast, southeast andsubcontinental Asia. Persons from the Middle East areincluded in the white NESB and white NAC groups inaccordance with international practice.

8. It is a crude estimate because ancestry responses areinflated by 16% over birthplace counts, given thatpeople were allowed up to two ancestries.

9. Only collection districts with more than 20 personsare included in all of these measurements.

10. The definition of the host community remains con-stant and therefore the percentage of minority groupsliving in the host community areas remains constant.However, when the minority groups are subdividedfurther their potential distribution across the minorityenclaves changes.

REFERENCESAl-Haj, M., 2002: Multiculturalism in deeply divided socie-

ties: the Israeli case.

International Journal of InterculturalRelations

26, 169–183.Bannerji, H., 2000: The paradox of diversity: the construc-

tion of a multicultural Canada and ‘women of color’.

Women’s Studies International Forum

23, 537–560.Beckett, D., 2001: Mainstreaming diversity in Australia’s

schools: problematizing the neo-liberal ‘third way’.

International Journal of Educational Research

35, 269–280.

Birrell, B. and Rapson, V., 2002: Two Australias: migrantsettlement at the end of the 20th century.

People andPlace 10, 11–25.

Burnley, I., 1994: Immigration, ancestry and residence inSydney. Australian Geographical Studies 32, 69–89.

DIMIA (Department of Immigration and Multicultural andIndigenous Affairs), 2003: Australian Multicultural PolicyRetrieved July, 2003 from <http://www.immi.gov.au/multicultural/australian/index.htm#into>.

Dixson, M., 1999: The Imaginary Australian: Anglo-Celtsand Identity — 1788 to the Present. University of NewSouth Wales Press, Sydney.

Healy, E. and Birrell, B., 2003: Metropolis divided: thepolitical dynamic of spatial inequality and migrant settle-ment in Sydney. People and Place 11, 65–87.

Iredale, R., 1997: Skills Transfer: International Migrationand Accreditation Issues. University of WollongongPress, Wollongong.

Jamrozik, A., Boland, C. and Urquhart, R., 1995: SocialChange and Cultural Transformation in Australia. Cam-bridge University Press, Cambridge, UK.

Jagorsky, P.A., 1997: Poverty and Place: Ghettos, Barriosand the American City. Russell Sage Foundation, New York.

Johnson, C., 2002: The dilemmas of ethnic privilege: Acomparison of construction of ‘British’, ‘English’ and‘Anglo-Celtic’ identity in contemporary British andAustralian political discourse. Ethnicities 2, 163–188.

Johnston, R., Forrest, J. and Poulsen, M., 2001: The geog-raphy of an EthniCity: residential segregation of birth-place and language groups in Sydney, 1996. HousingStudies 16, 569–594.

Johnston, R. and Poulsen, M., 2000: The Ghetto Model andethnic concentration in Australian Cities. Urban Geogra-phy 21, 26–44.

Johnston, R., Poulsen, M. and Forrest, J., 2003a: Ethnicresidential concentration and a ‘new spatial order?’:exploratory analyses of four United States metropolitanareas, 1980–2000. International Journal of PopulationGeography 9, 39–56.

Johnston, R., Poulsen, M. and Forrest, J., 2003b: And didthe walls come tumbling down? Ethnic residential segre-gation in four US metropolitan areas 1980–2000. UrbanGeography 24, 560–581.

Johnston, R., Poulsen, M. and Forrest, J., 2003c: The ethnicgeography of New Zealand: a decade of growth andchange, 1991–2001. Asia Pacific Viewpoint 44, 109–130.

Johnston, R., Wilson, D. and Burgess, S., 2004: Schoolsegregation in multi-ethnic England. Ethnicities 5, 237–265.

Jones, F.L., 1991: Ancestry Groups in Australia: A DescriptiveOverview. Centre for Multicultural Studies, University ofWollongong for Office of Multicultural Affairs, Canberra.

Jones, F.L., 1996: Ethnic enclaves; a transitory phenomenon.People and Place 4, 32–33.

Jupp, J., 1989: The Challenge of Diversity: Policy Optionsfor a Multicultural Australia. Australian GovernmentPublishing Service, Canberra.

Khoo, S-E. and Birrell, B., 2002: The progress of youngpeople of migrant origin in Australia. People and Place10, 30–44.

Kunz, C., 2003: 2001 Census: Ancestry — first and secondgeneration Australians. Australian Bureau of StatisticsCensus Paper, No. 03/01a.

Is Sydney a Divided City Ethnically? 377

© Institute of Australian Geographers 2004

Kunz, C. and Costello, L., 2003: 2001 Census: Ancestry —Detailed Paper. Australian Bureau of Statistics CensusPaper, No. 03/01b.

Ley, D. and Murphy, P., 2001: Immigration in gateway cit-ies: Sydney and Vancouver in comparative perspective.Progress in Planning 55, 119–194.

MacLaughlin, J., 1998: Racism, ethnicity and multiculturalismin contemporary Europe: a review essay. Political Geog-raphy 17, 1013–1024.

Marden, P. and Mercer, D., 1998: Locating strangers: multi-culturalism, citizenship and nationhood in Australia.Political Geography 12, 939–958.

Marcuse, P. and van Kempen, R., 2000: Introduction. InMarcuse, P. and van Kempen, R. (eds) Globalizing Cities:A New Spatial Order. Blackwell, Oxford, 1–21.

Massey, D.S., 1985: Ethnic residential segregation: a theo-retical synthesis and empirical review. Sociology andSocial Research 69, 315– 350.

Morrissey, M., Dibden, M. and Mitchell, C., 1992: Immigra-tion and Industry Restructuring in the Illawarra. Bureauof Immigration Research, Australian Government Publish-ing Service, Canberra.

NMAC (National Multicultural Advisory Council), 1999:Multiculturalism: The Way Forward. Australian Govern-ment Publishing Service, Canberra.

Peach, C., 1996: Does Britain have ghettos? Transactionsof the Institute of British Geographers NS 22, 216–235.

Philpott, T.L., 1978: The Slum and the Ghetto: Neighborhood

Deterioration and Middle-class Reform, Chicago 1880–1930. Oxford University Press, New York.

Portes, A. and Rumbaut, R., 2001: Legacies, the Story of theImmigrant Second Generation. University of CaliforniaPress, Berkeley.

Poulsen, M., Forrest, J. and Johnston R., 2002b: Frommodern to post-modern? Contemporary ethnic residentialsegregation in four US metropolitan areas. Cities 19, 161–172.

Poulsen, M., Johnston, R. and Forrest, J., 2001a: Sydney’sethnic geography: new approaches to analysing patternsof residential concentration. Australian Geographer 32,149–162.

Poulsen, M., Johnston, R. and Forrest, J., 2001b: Intraurbanethnic enclaves: introducing a knowledge-based classifica-tion method. Environment and Planning A 33, 2071–2082.

Poulsen, M., Johnston, R. and Forrest, J., 2002a: Pluralcities and ethnic enclaves: introducing a measurementprocedure for comparative study. International Journal ofUrban and Regional Research 26, 229–243.

Smolicz, J.J., 1995: The emergence of Australia as a multi-cultural nation: an international perspective. Journal ofIntercultural Studies 16, 2–24.

Williams, L. and Batrouney, T., 1998: Immigrants andpoverty. In Fincher, R. and Nieuwenhuysen, J., (eds)Australian Poverty: Then and Now. Melbourne UniversityPress, Melbourne, 258–275.

Zang, X. and Hassan, R., 1996: Residential choices of immi-grants in Australia. International Migration 34, 567–582.