SOME GEOGRAPHIC DIMENSIONS OF BEING WORK-RICH AND WORK-POOR: CHANGES BETWEEN 1986 AND 1996(1).

Consider for a moment a neighbourhood in which most working-age women are not in paid jobs. This may conjure up a picture of tidy homes, children at play and gossip. Now think of a neighbourhood in which most men are jobless. The picture is more sinister. Areas of male idleness are considered, and often are, places of deterioration, disorder and danger. Non-working women are mothers; non-working men, a blight. (The Economist, 28 September, 1996:27)

INTRODUCTION

A geographic dimension is increasingly considered to be of importance in research on the distribution of paid work, income, wealth, and well-being within industrialised societies (e.g. Crampton et al. 1997, Cutler and Glaeser 1995, Gregory and Hunter 1995, Morrison 1993a). Such research recognises that in areas where a high proportion of people, and particularly working-age men, are not in paid work this may impact negatively on a range of economic, social and health indicators for the residents in these communities. For example, having high unemployment in a community can be associated with lower levels of provision and quality of services such as shops, transport, health facilities, schooling, and housing. Van Kempen (1997) even suggests that in "poverty pockets" the quality of welfare delivery could be lower than in other areas. There may also be "neighbourhood" effects in terms of negative role models and social norms (e.g. Case and Katz 1991).(2) Conversely, research also suggests that communities which are cohesive through a high rate of participation in community-based activities are also those which are economically prosperous (Putnam 1993).(3)

People with certain similar characteristics tend to cluster together in particular geographic areas. Various reasons have been proposed as to why this might occur. These include the provision of low-cost public housing bringing together low-income people; inner-city crime forcing middle-class residents out to the suburbs; or only high-income people being able to afford the prime geographic sites (e.g. Coleman 1997, Mills and Lubuele 1997). Research in the United States shows that clustering by some characteristics, such as religion or ethnicity, can result in a relatively wide mix of employment and income-earning ability in a community (Wilson 1987). However, the same research suggests that in black, inner-city ghettos the well educated have tended to move out in recent years.(4) With educational attainment increasingly representing a key factor in determining labour market outcomes for individuals, clustering by educational attainment is far more likely to lead to a homogeneous community in terms of employment patterns and income. Again in the United States, Herrnstein and Murray (1996:xxi-xxii) argue that those with good jobs, education and incomes "... gravitate to one another, increasingly enabled by their affluence and by technology to work together and live in one another's company -- and in isolation from everybody else". Similarly, the Brookings Institution (1997) describes how "gated" residential communities with private security guards, private gardeners, and other private facilities are becoming more common in the United States for those in the upper-income brackets. Linking this to changes in employment, Reich (1993) describes how highly educated people with "symbolic-analytical" type jobs are increasingly well placed in both local and international labour markets. Reich also discusses the manner in which this group "share" their income:

   In allocating personal income, the symbolic analyst has shown no lack of
   willingness to engage in collective investment. But increasingly, the
   public goods that result are shared only with other symbolic analysts.
   Symbolic analysts take on the responsibilities of citizenship, but the
   communities they create are composed only of citizens with incomes close to
   their own. In this way, symbolic analysts are quietly seceding from the
   large and diverse publics of America into homogeneous enclaves, within
   which their earnings need not be redistributed to people less fortunate
   than themselves. (p.268)

Reich goes on to suggest that as the highly educated, high-income groups seek tax cuts they effectively "... withdraw their dollars from the support of public spaces shared by all and dedicate the savings to private spaces they share with other symbolic analysts." There is also a range of other policy implications that arise as a result of similar people grouping together. For example, there is a concern in both New Zealand and overseas about the quality of schooling in areas where low-income people are concentrated.

This paper explores some geographic dimensions of two factors that strongly influence the distribution of paid work, income, wealth, and ultimately, well-being. These are participation in paid work and, linked to this, educational attainment. In particular, the paper examines whether, in parallel to the growth of work-rich and work-poor individuals and families (Callister 1998), some geographic areas are also becoming work poor or work rich. Finally, in order to assess the possible relationship of paid work to a wider concept of well-being, a New Zealand index of deprivation is compared with the geographically based employment data.

DEFINITIONS

Geographic Areas

In New Zealand, there has recently been some interest in building stronger communities. This includes the concept of delivering more services and developing employment initiatives at a community level. Unfortunately, the concept of a community is often unclear. In a recent "thinkpiece" the Department of Internal Affairs (1997) acknowledges that the term "community" has a variety of meanings, with only one of them being the place in which people live. For this research, it would be useful if there were geographic areas that were not only places to live and interact socially, but were also geographically defined labour markets. In Wellington, Morrison (1995) found that a significant proportion of people worked locally. However, he also found that people in areas that were "job-poor", in general, had to travel the greatest distances when in paid work. In some other communities, such as rural areas, people often live where they work such as on a family farm. However, even in traditional farming areas many people now commute to off-farm jobs. In addition, improved communications and globalisation of work mean that a workplace could also be a home, but with the employer on the other side of the world rather than across town. While these complexities are important in understanding the changing dynamics of communities, in this research I focus on the areas in which people live.

The smallest geographic area used by Statistics New Zealand is the meshblock. There are three disadvantages associated with using meshblocks. Firstly, the small number of people in each meshblock can present confidentiality problems, thereby limiting the analysis. Secondly, as meshblocks often represent only a couple of streets they may be too small to create "neighbourhood effects". However, most importantly, meshblocks have coded names which are not generally recognisable. Having recognisable names assists policy makers and the general public locate and consider the characteristics of an area.

The first aggregation of meshblocks is an area unit. Area units are easily identified as each area unit must be a single geographic entity with a unique name referring to a geographical feature. For this research, work patterns within 1,636 area units were analysed (see appendix for a more detailed discussion of issues involved in selecting the areas to be analysed).

Work Rich and Work Poor Area Units

In a previous paper, the concept of work-rich and work-poor prime-age individuals and families was explored (Callister 1998). Prime-aged people were defined as being in the 25-59 age group. There are, however, various possible measures for determining the work status of a geographic area (see appendix for a brief discussion). In this study a new approach was taken. First, the hours of paid work were added together for every prime-age person in each area unit. This measure of total hours of paid work in each unit was then divided by the total population, whether in paid work or not, in the target age group in each area. This gave an average of hours of involvement in paid work per person per week across the total prime-age population.(5) This calculation was carried out using data from the 1986, 1991 and 1996 censuses to provide a time series. This type of measure controls for differences in population in each area unit, although there are still some problems with a potentially changing age distribution within the prime-age group within each area.

SOME GEOGRAPHIC CHANGES IN HOURS OF PAID WORK

As discussed in Callister (1998), there were major changes in the population, participation in paid work, and hours worked for prime-aged women and men between 1986 and 1991, and again between 1991 and 1996. Overall, the changes at a national level meant that in 1986 there were 32.7 hours worked per week for each prime-age person, in 1991 this had declined to 29.3 hours, and then in 1996 increased back to 32 hours only slightly under the 1986 figure.

Table 1 shows the proportion of communities that might be considered work rich or work poor using five cut-off points. It firstly shows that, like poverty lines, there can be great sensitivity around the choice of cut-off point for work-poor area units. However, the table does show a rise in the proportion of areas that were work-poor, in each of the chosen measures, between 1986 and 1996. In contrast, there was little or no rise in the proportion of areas that were work rich. It also shows that the loss of paid work between 1986 and 1991 had an impact on a significant number of areas within New Zealand.

Table 1 Percentage of Area Units by Average Hours of Paid Work: Prime-Age People

Average hours per person          1986     1991     1996

Under 20 hours (work poor 1)       0.4      4.1      1.4
Under 25 hours (work poor 2)       1.7     15.2      7.4
Under 30 hours (work poor 3)      15.3     52.0     27.4
35 or more hours (work rich 1)    24.6     11.9     25.4
40 or more hours (work rich 2)     6.2      2.8      6.2

Source: Census of Population and Dwellings, Statistics New Zealand.

However, Table 1 provides no idea of whether it was the same areas that were work-rich or work-poor in each census period.

Figure 1 shows the distribution of area units by average hours for prime-age people in 1986 as well as the average hours for the same areas ten years later. The data for each area unit are presented in ascending order of 1986 average number of hours of paid work (the white line), and the 1996 data for each area unit are plotted in black.

[Figure 1 ILLUSTRATION OMITTED]

The graph shows that there were considerable changes in the average hours per person in many individual area units. However, the extremely large shifts tended to be in very small area units.(6) Overall, in this period of initial major job loss followed by strong employment growth the average hours worked per person per week declined in 57% of area units.

There are many dimensions to these changes, including some differences between urban and rural areas. To illustrate this, Figure 2 restricts the analysis to changes within the Auckland urban area between 1986 and 1996. When compared with Figure 1 the graph shows that there were far fewer areas which had very high average hours of paid work in 1986. In addition, there was a decline through to 1996 in average hours per person of those ten areas with the highest hours in 1986. In some of these areas, this simply reflects a process of urbanisation in which farmland, where people tend to work long hours, was subdivided. However, of more interest to policy makers was the decline in average hours of paid work between 1986 and 1996 in many of those areas which were already work poor in 1986. These areas include communities well known to researchers and social service agencies including Otara East, Otara West and Glen Innes East.(7) The one community within those ten with the lowest average hours of paid work in 1986 that did show a marked increase in average hours was Point Chevalier South. This primarily reflects "gentrification" which, in turn, was linked to changes in government housing policy. Former state houses, which were in a prime location close to the harbour and city centre, were sold to higher-income people. This gentrification process can be seen in the proportion of the prime-age population in this area who had a degree or higher qualification. In 1986, 5% of the prime-age population in Point Chevalier South had a degree or higher, which was below the national average. However, by 1996, in this area unit the proportion of the prime-age population with a degree or higher had risen to over 16%, over the national average. This example, and the rural subdivision example, illustrate some ways in which the work status of a community can change over time.

[Figure 2 ILLUSTRATION OMITTED]

In order to provide a further New Zealand-wide analysis, area units were grouped into deciles by their average hours of paid work per person in each census. Given the concern about the lack of paid work in many area units, there is a particular focus on the status and characteristics of a group of area units that might be considered to be work poor. These are the area units in the bottom decile.(8) In contrast, work-rich area units are seen as those in the top decile of hours of paid work.

Of those area units that were in the bottom decile of hours in 1986, 60% were still in the bottom decile in 1991. The movement out of the bottom decile was not as strong in the next five years, despite this being a period of significant overall employment growth. Of those area units that were in the bottom decile of hours in 1991, 78% were again in the bottom decile in 1996.

Overall, 52% of those area units in the bottom decile in 1986 were still in the bottom decile in both 1991 and 1996. In addition, of those in the bottom decile in 1986, 82% were in the bottom two deciles in 1996, a further 7% in the third to bottom decile, and 4% had moved into the forth to bottom decile. Only two very small rural areas rose from the bottom decile in 1986 to reach the third to top decile by 1996. Also noteworthy is the fact that the average hours of paid work per person increased between 1986 and 1996 in only 32% of the area units in the bottom decile in 1986.(9)

These data lend some support to the Australian work of Gregory and Hunter (1995), which found that in particular areas labour market disadvantage continues over a long period. But the data also suggest that there is some, albeit limited, mobility of area units in terms of average hours of paid work undertaken.

Of critical importance, however, is whether large groups of people remain for long periods in work-poor areas. Some US research suggests that there is considerable movement in and out of poor urban areas, even amongst those people who are persistently poor (Gramlich et al. 1992).(10) Gregory and Hunter gave no indication as to whether the same people were in the disadvantaged communities in each time period. In New Zealand, it may be that a high proportion of people spend a short time in work-poor area units and move out when their employment prospects, education or income improve. Or, it may be that many people are mobile but simply move from one work-poor area to another work-poor area. Such mobility may be due to insecure housing arrangements rather than for more positive reasons. These New Zealand data do not directly shed light on these issues. However, in some area units, such as Otara East, North and South, which were work-poor in each census, there was a high concentration of people from Pacific Island ethnic groups in each time period. If there had been a considerable level of mobility this concentration might be expected to reduce (although this might be masked by inward migration), and that there would be far more Pacific Island people living in work-rich areas. In addition, in a study of the long term unemployed, Parker (1997) notes that around 40% of the respondents stated that they could not move to areas with more jobs. These data would suggest that there may only be limited mobility for at least some groups of people.

Finally, while the mobility of work-poor adults may be important in terms of their finding paid work, it can have a negative impact on learning opportunities for their children. For instance, there are already concerns about the high roll turnover in many South Auckland schools. This impacts, in turn, on the educational outcomes, and ultimately job prospects, of the next generation.

THE LOCATION OF WORK-RICH AND WORK-POOR AREAS

The data indicate that while many work-poor areas were to be found in large urban areas such as South Auckland or the Porirua Basin near Wellington, they were also to be found throughout rural and urban New Zealand. However, in the rural areas it was the small rural service centres rather than the farming areas that were particularly prone to being work poor. In 1996, rural centres represented 8.2% of the area units studied, but made up 14.1% of work-poor area units. Some small towns, such as Patea and Shannon, were also work-poor in each census. Most urban areas had at least one area unit that was work-poor over a long period. Examples include Otangarei in Whangarei, Fordlands in Rotorua, Orini in Whakatane, Lower Aramoho in Wanganui, Taita North in Lower Hutt, Addington in Christchurch and South Dunedin.

In rural New Zealand, while some work-poor areas exist, a significant proportion of work-rich area units were also to be found. In each of the censuses, around 70% of the work-rich area units were rural, with rural centers representing a further 5-6% in each census.(11) A small number of the work-rich rural areas, as well as those within small towns or urban areas, were primarily tourist destinations rather than being service centres for farmers. Examples were Fox Glacier, Mt Cook, and Queenstown. However, the majority of the rural work-rich areas were farming area units. This included dairy farming areas, where long hours of paid (and unpaid) work are common, especially at the time of the year the census was taken. These work-rich rural areas are not necessarily income-rich, with long hours often worked in family businesses, regardless of the economic cycle.(12) However, there were some area units in major urban areas which were work rich. Unlike the work-rich rural area units, these urban area units tended to contain people with above-average levels of education. For example, in 1996, in the Wellington area, there were five area units that were in the top decile of work hours nationally. These were Karaka Bay-Worser Bay, Wadestown, Roseneath, Churton Park, and Grenada. In the first three, there were particularly high levels of education amongst the population. Overall, of those area units that were work-rich in 1986, just under two-thirds were again in this top decile in both 1991 and 1996. These long-term work-rich areas tended to be the farming communities.

GENDER ISSUES

Perhaps, as the quote from The Economist at the beginning of this paper suggests, the "problem" communities are only those where there is little paid work amongst prime-age men. The analysis of average hours in area units disguises some major changes in women's and men's paid work hours over the period 1986 to 1996. Previous research (Callister 1998) showed that, over this period, participation in paid work increased for prime-age women, as did the hours worked for many of these women. For prime-age men participation declined, as did the hours worked for many men still in paid work. These patterns show up strongly when trends within area units are examined. However, Table 2 reaffirms that women, overall, have both a lower participation in paid work and, when they are in it, tend to work shorter hours than men.

Table 2 Percentage of Area Units Where the Average Hours of Paid Work for all Prime-Aged Women or Men Was Less than 30 Hours Per Week

        1986   1991   1996

Women   97.7   98.2   91.0
Men      0.8    9.5    6.7

Source: Census of Population and Dwellings, Statistics New Zealand.

While job losses for men were widespread between 1986 and 1991, there were some very major declines in men's average hours in some area units. For example, in six area units, Tokomaru Bay, Kawau, Whangarei Central, Kaingaroa Forest, West Invercargill, and Moerewa, men's average hours across the area unit declined by 20 hours or more. These very large declines can usually be traced to a particular event, such as the closing of a freezing works in the situation of Moerewa and the restructuring of state forestry in the Kaingaroa Forest area unit.

In order to gain a further insight into the possible links between men's and women's patterns of paid work, Table 3 provides a breakdown of the growth or loss of hours of paid work in area units by gender between 1986 and 1996.

Table 3 Area Units by Change in Average Hours for Prime-Age Men and Women: 1986-1996

                                      Male hours   Male hours
                                        declined   steady or
                                                   increased

Female hours steady or increased   1,243 (76.0%)    158 (9.7%)
Female hours declined                218 (13.3%)     17 (1.0%)

Source: Census of Population and Dwellings, Statistics New Zealand.

In this ten-year period, in nearly 86% of area units, average female hours remained steady or increased. In just under 11% of area units average male hours stayed steady or increased. One of the effects of these changes is that by the end of the ten-year period there was a closer relationship between men's average hours and women's average hours of paid work within the area units studied. Running a simple linear regression across the data in 1996 produced an [r.sup.2] of 0.41 as against 0.15 in 1986. This suggests that, increasingly, when men face major difficulties in a particular labour market, then opportunities for employment by women are also constrained.(13) Other area units tend to have high hours of paid work for both men and women. In the situation of couples, these trends are shown up in the shift from a mixed-work, male primary-income-earner model to either a work-rich, two earner, model or a work-poor, no earner, model (Callister 1998).

The following table provides four specific examples of area units. Otara East is a South Auckland urban area that was in the bottom (most work-poor) decile in all three censuses examined in this paper. Hauraki Plains is a North Island farming area that was in the top decile of hours in each census. Wadestown is an education-rich, high-income, central Wellington suburb and an area that moved into the most work-rich decile in 1996.(14) Finally, Moerewa, a small rural town in Northland, is shown. Moerewa moved into the bottom decile of work in 1991 and was again in this decile in 1996. The situation of Moerewa highlights the difficulty experienced by small towns in providing new employment opportunities once a key industry has closed.

Table 4 Average Hours for Prime-Age Men and Women in Examples of Work-Rich and Work-Poor Area Units: 1986-1996

Otara East         1986   1991   1996

  Male hours       33.9   19.9   29.0
  Female hours     16.9   11.2   16.9

Hauraki Plains
  Male hours       57.4   50.8   54.7
  Female hours     28.4   26.9   27.4

Moerewa
  Male hours       41.5   21.3   22.7
  Female hours     18.3   12.6   15.1

Wadestown
  Male hours       44.6   43.7   46.2
  Female hours     26.2   27.5   31.2

Source: Census of Population and Dwellings, Statistics New Zealand.

Comparing the data for Otara East and Wadestown shows a trend also noted by Gregory and Hunter in Australia. That is, by the mid-1990s, while within each area men were involved in more hours of paid work than women, in some of the work-rich area units women's labour market position is now better than that of men in work-poor area units. This reflects both educational and ethnic differences, with well-educated Pakeha women increasingly better placed in the labour market than poorly educated Maori and Pacific Islands men.

DO WORK-POOR COUPLES LIVE IN WORK-POOR AREAS?

While it is useful analysing women's and men's work patterns separately many live together as couples. It would then be expected that work-poor couples would tend to live in work-poor areas. However, there are two issues, which are both of potential interest to policy makers, that need to be considered when analysing couples. First, there is overseas evidence that heterosexual-couple households are less common in areas where there is an excess of either women or men in the area, and/or where there is a high proportion of men who are not in paid work (e.g. Wilson 1987).(15) Second, if couples do form there is then the question as to what proportion are then work rich or work poor(16) Table 5 explores these issues.

Table 5 Some Family Characteristics of Area Units by Average Hours of Paid Work per Prime-Age Person, 1996 data

Deciles     No of        Ratio of     Ratio of    No of
sorted by   prime-aged   prime-aged   women to    prime-aged
1996 data   people       women to     men in      couples(*)
                         men          paid work

Work poor
Decile 1       137,442   1.04        1.75           33,303

Decile 2       183,960   1.08        1.48           50,522

Decile 3       178,698   1.06        1.37           50,880

Decile 4       202,689   1.06        1.31           59,546

Decile 5       190,272   1.05        1.27           59,785

Decile 6       206,823   1.04        1.23           65,831

Decile 7       193,932   1.04        1.20           62,680

Decile 8       150,258   1.02        1.17           49,883

Decile 9       139,599   1.00        1.14           46,816

Decile 10      109,842   0.91        1.03           39,398
Work rich

Total        1,693,515   1.04        1.28          518,644

Deciles     Couples as   No of        Work-poor
sorted by   % of         prime-aged   couples as
1996 data   prime-age    work-poor    a % of all
            population   couples      couples

Work poor
Decile 1     24.2          6,064         18.2

Decile 2     27.5          5,745         11.4

Decile 3     28.5          4,499          8.8

Decile 4     29.4          4,260          7.2

Decile 5     31.4          3,921          6.6

Decile 6     31.8          3,489          5.3

Decile 7     32.3          2,915          4.7

Decile 8     33.2          1,888          3.8

Decile 9     33.5          1,617          3.5

Decile 10    35.9          1,364          3.5
Work rich

Total        30.6         35,762          6.9

Source: Census of Population and Dwellings, Statistics New Zealand.

First, there seems no obvious relationship between the ratio of women to men and the proportion of couples in a population. However, as predicted, the table shows some tendency for an increased proportion of couples in the population as the hours of paid work in area units increase and, at the same time, as the ratio of all women to men in paid work decreases. While there are dangers in assuming any direct causation, these data are consistent with theories that lack of paid work for both men and women affects the formation or stability of couples. If so, then in childrearing situations, sole parenthood (albeit mainly sole motherhood) can be expected to be more common in work-poor areas. The lack of employment in these areas or nearby labour markets can constrain attempts to encourage sole parents into paid work. Likewise, an inadequate supply of affordable, high quality childcare in these work-poor areas could provide a supply-side restraint to participation of these parents in paid work.

It is also not surprising that a declining proportion of couples were work poor as the amount of paid work undertaken within the community increased. It may simply be that couples need both partners to be in paid work in order to live in work-rich areas, or that work-poor couples can only afford to live in work-poor areas. However, the high proportion of couples who were work poor in work-poor areas is also consistent with the idea that it is a lack of job opportunities in local (or nearby) labour markets, rather than "fixed-gender roles" for men and women in paid and unpaid work and/or the design of welfare benefits, which is the critical factor in the growth of this family type. However, given that there were still work-poor couples in work-rich areas, this suggests that there are some supply-side factors which could also be importune.(17)

SOME REASONS FOR AREA UNITS BEING WORK-RICH OR WORK-POOR

There are likely to be a wide range of reasons as to why an area might be work rich or work poor or attract people who are work rich or work poor. Many, such as the number and viability of local employers, the level of crime in an area, the topography, or the quality of local services, such as schools, are either difficult to measure or the data can only be compared with census data by integrating a number of databases.

Simply based on census data, the work-poor area units in each year were a somewhat mixed group. For example, included in the long-term group of work-poor area units were a small number of sites of either major prisons, psychiatric institutions, or educational institutions where large groups of people are not in the labour force (including partners who might move there to be near their mates). These areas included Paremoremo East, Rangipo, and Tokanui, and Massey University. Some work-poor areas can be traced to a specific industry collapse such as Patea where the freezing industry collapsed in the early 1980s. Others, such as Foxton Beach, have been in a long-term economic decline.

There are some work-poor urban areas that are well known to researchers and policy makers, such as those in South Auckland, Porirua, and Lower Hutt. In these latter area units, the housing market and, in particular, the concentration of state housing appears to be an important factor in the concentration of work-poor people and families (Morrison 1995). Morrison notes that much of the state housing was built to service manufacturing centres. When labour demand for manufacturing declined dramatically in the 1980s, these people were then isolated from areas where jobs were increasing. Yet policies such as providing housing assistance for private rental dwellings should be reducing this concentration.

The housing market could also be important in some other situations because house ownership in a depressed area can make moving to areas of job growth more difficult (The Economist 1997). This will be particularly problematic if there are significant price differences between areas. It may also be that some work-poor individuals have moved to small rural towns in order to find low-cost accommodation, with one result being that they are further isolated from the job market. In specific relation to work-poor Maori, Mare (1995) suggests that some geographic immobility may be due to attachment to established iwi areas.

In terms of work-rich areas, issues such as availability of jobs and the type of job are clearly important. However, other factors, such as the provision of high quality childcare and, for those with sufficient income, other household support services such as cleaners, are likely to have a strong influence on hours of paid work, particularly for women(18)

At an individual level, education is an important factor in labour market participation and earnings. Linked to this, ethnicity is also an important variable in the labour market. Both these factors seem to be an important variable in clustering behaviour. The following table groups area units in deciles by paid work hours in 1996. The characteristics of these groups are then shown in terms of the proportion of prime-age people with degrees or higher, the percentage with school or tertiary qualifications, and the percentage who were from Maori or Pacific Islands ethnic groups. In addition, as a crude test of the link between community participation and economic activity, the proportion of people who said they undertook community work is shown.(19)

For education, for deciles 1-8, there is a clear link between the proportion of people who have formal qualifications, in particular degrees or higher, and the hours of paid work in an area unit. The slightly lower education rates in decile 9, but particularly the low rates in decile 10, are associated with the over-representation of farming areas. A lack of formal qualifications has not historically been such a barrier to participation in paid work on family farms. However, this is likely to change as farming increasingly comes to be seen primarily as a business rather than as a way of life. It is also clear that Maori and Pacific Island peoples are significantly over-represented in work-poor areas.

Although the differences are not great, there was an increasing proportion of people who undertook community work as the average hours of paid work in an area increased. This seems to support the idea that areas where people undertake a high level of voluntary work are "cohesive" and therefore more prosperous. However, there could be many reasons for the relationship. For example, the higher rate of community work, particularly in the rural area units, may simply reflect a lack of commercial services in some areas, which then need to be provided by unpaid workers. The lower rate in some work-poor areas may also reflect a shortage of physical capital for facilities, such as halls or buildings suitable for playgroups, which assist in developing strong community networks.

As with most grouped measures, there are some major exceptions to these broad patterns. For example, the Waiheke Island area unit, which was in the bottom decile of paid work not only in 1996 but also in the previous two censuses, had nearly 12% of the prime-aged population with degrees or higher in 1996.

The data in Table 6 also indicate that education tends to be concentrated in particular area units, and people with tertiary qualifications, and particularly those with degrees or higher, have some tendency to cluster together in particular areas.(20) It will be those people with degrees or higher who tend to be the "symbolic analysts" that Reich (1993) describes as clustering together in the United States. These are also the people who, according to Reich, increasingly are the generators of ideas, wealth and, ultimately, jobs in an economy.

Table 6 Characteristics of Area Units by Average Hours of Paid Work per Prime-Age Person, 1996 data

Deciles sorted by       % of prime-   % of prime-aged
1996 data               aged people   people with
                        with degree   post school
                        or higher     qualifications
                                      or higher

Decile 1 - Work-poor        4.4          24.0
Decile 2                    6.2          30.2
Decile 3                    7.8          33.5
Decile 4                    9.9          37.4
Decile 5                   11.1          40.1
Decile 6                   12.2          42.0
Decile 7                   14.9          45.2
Decile 8                   17.3          47.9
Decile 9                   16.4          45.9
Decile 10 - Work-rich       7.5          35.2
Total                      10.9          38.4

Deciles sorted by       % of prime-aged   % of prime-
1996 data               people of Maori   aged people
                        or Pacific        who undertook
                        Islands           community
                        ethnic groups     work

Decile 1 - Work-poor        47.7            38.6
Decile 2                    27.9            41.3
Decile 3                    20.8            42.4
Decile 4                    17.2            42.1
Decile 5                    13.4            43.2
Decile 6                    10.2            43.3
Decile 7                     9.2            43.1
Decile 8                     8.8            44.4
Decile 9                     8.9            44.3
Decile 10 - Work-rich        7.9            48.1
Total                       16.8            42.9

Source: Census of Population and Dwellings, Statistics New Zealand.

As an alternative measure to examining the proportions of various qualifications within each area unit, we can look at the distribution of qualifications overall. In 1996, 44% of all prime-age people with degrees or higher qualifications lived in 10% of the area units (the top decile). At the other end of the spectrum, the bottom 10% of communities contained around 0.5% of those prime-age people with degrees. When a wider range of qualifications is considered, a similar but less extreme pattern shows up. Again for 1996, for all post-school qualifications, the top 10% of communities contained 27% of those with such qualifications. The bottom 10% contained only 1.1% of people with post-school qualifications. This type of measure could be problematic because areas units contain different numbers of people, yet further testing of the data showed little relationship between the size of each area unit and the percentage of prime-age people with a degree. However, people with degrees were more likely to live in area units within urban areas. In 1996, in the combined grouping of rural areas and rural centres, 6.1% of prime-age people had degrees, while in non-rural areas this nearly doubled to 11.8%.

Again, this analysis of education is a static view. Of equal interest is whether the increasing participation in tertiary education is resulting in education-poor area units improving their position. Unfortunately, changes in the census education question make comparisons for all tertiary education difficult. Therefore, in the following table, area units are grouped into deciles by the percentage of prime-age people with a degree or higher. The change in the percentage of people with degrees or higher in these same communities between 1986 and 1996 was then calculated.

The table shows that while all deciles gained from the increase in university-educated people, the deciles which gained the most from the increase in the proportion of prime-age people with degrees or higher between 1986 and 1996 were ones which already had the highest proportion of people with degrees in 1986. This means that either people in education-poor area units are less likely to obtain degrees or, if they do, they then tend to move to area units which already have a higher proportion of people with this type of qualification. Either way, it suggests that many of the gains from increased public and private spending on university education may not be directly spreading into the area units which could most benefit from an increase in the proportion of well qualified people.

Table 7 Percentage of Prime-Age People with a Degree or Higher by Area Units, 1986 and 1996

                                      % of
                                 prime-age
                               people with
                               a degree or
                                higher(*)       Absolute

Deciles sorted by 1986 data    1986   1996   change in %

Decile 1 - Education poor       1.1    2.4           1.3
Decile 2                        2.4    3.8           1.4
Decile 3                        3.2    4.6           1.5
Decile 4                        3.8    5.3           1.5
Decile 5                        4.6    6.7           2.1
Decile 6                        5.4    7.8           2.4
Decile 7                        6.5    8.8           2.4
Decile 8                        8.1   11.2           3.0
Decile 9                       11.3   16.5           5.2
Decile 10 - Education rich     21.7   29.6           7.9
Total                           7.4   10.9           3.5

Source: Census of Population and Dwellings, Statistics New Zealand.

(*) This is an average of the percentages in each area unit

It also suggests that if the link between education, paid work and income generation becomes even stronger, then there is the potential for even greater disparity in both work and income in the future between work-poor and work-rich area units. This is likely as many researchers suggest we are moving into an era where the ability to handle and analyse complex information is becoming a key skill (e.g. Castells 1996).

ARE WORK-POOR AREAS DEPRIVED?

A final question of relevance to social policy is whether work-poor communities are deprived. It could be that some communities with low levels of paid work amongst prime-age people are simply areas where people are choosing a low income but a high quality lifestyle, with Waiheke Island perhaps an example. There could also be areas where a significant number of people have retired early with sufficient wealth to ensure a good quality of life, or where couples choose to live on one income. There is much debate over quality of life or well-being measures. However, one New Zealand measure that can easily be applied to area units is the Index of Deprivation. This was initially developed using 1991 census data (Crampton et al. 1997) but has now been updated to 1996. This index is based on a range of variables including unemployment, quality of housing, access to a car, access to a telephone and household income levels.

A regression analysis shows a reasonably close relationship between the 1996 index of deprivation and the average paid work hours of prime-aged people in area units in 1996. For total average hours of paid work there was an [r.sup.2] of 0.55, while for the hours of men only this rose to 0.57.(21) However, perhaps of more importance is the matching at the low end of hours of paid work. When the work-poor decile (164 area units) was compared with the decile of area units with the highest level of deprivation, 120 area units (or 73%) fell within both criteria. Examples of area units that did not meet the double criteria were Waiheke Island and Great Barrier Island. While further research is needed in terms of linking patterns of paid work to well-being, these data would suggest that, currently, in New Zealand, availability of paid work in a geographic area is a major factor in determining the well-being of individuals in that area.

CONCLUSION

While perhaps not exhibiting the extremes seen in the United States, New Zealand does have enclaves of work-rich and work-poor people and, linked to this, areas where there are high and low concentrations of formal educational qualifications. This, along with the index of deprivation data, suggest there is an important geographic dimension to the distribution of income, wealth and, ultimately, well-being in New Zealand.

On various measures, the proportion of work-poor areas increased between 1986 and 1996. This was mainly caused by the loss of paid work amongst prime-age men, particularly amongst those with few formal qualifications. Although women's hours of paid work increased in many areas, this was not sufficient to compensate for the loss of work by men in these same areas.

Work-poor area units can be found throughout New Zealand. Key features of work-poor areas are both a low level of formal qualifications and a high proportion of residents who are from Maori or, within urban areas, from Pacific Islands groups. Within the large cities, work-poor area units are most likely to be found within traditional state housing areas, such as Otara in Auckland and Porirua in Wellington. However, in the mid-1990s, small rural centres were over-represented amongst work-poor communities. This latter trend appears to reflect the general running down of small rural service centres.

In all the periods studied, a high proportion of the work-rich area units were communities in which farming or tourism were key industries. However, there were also some workrich urban areas. Unlike work-rich rural areas, the urban areas tend to have a relatively high proportion of people with formal qualifications. In these urban areas, there has been a particularly strong increase in the participation and hours of paid employment of prime-age women. In these work-rich areas, prime-age women are more likely to be in couples with male partners who also had high rates of participation in paid work. Historically, some of these urban work-rich, and education-rich, areas would have been the home of "traditional" mixed-work childrearing families, with the man in full-time paid work and the woman either working part-time or not in paid work. Increasingly, however, in these areas couples (both with and without children) are becoming dual-job and, sometimes, dual-career families. It is likely that the average hours of paid work in these education-rich urban communities will keep rising as women increase both their labour force participation and their hours of paid work. This can be expected to equate with increasing incomes for these families and, potentially, increased financial resources for the communities in which they live.

The data also suggest that couple formation or stability may be linked to patterns of paid work, with couple families more prevalent in areas where paid work is plentiful. This suggests that labour markets, housing markets and "marriage markets" are closely interlinked. However, policy makers often treat them separately.

The reasons for the development of work-poor areas are complex and need further exploration. However, they include a lack of labour demand, low levels of formal qualifications in most of these areas and, primarily in large urban areas, concentrations of relatively low-cost housing which bring people with educational and other potential disadvantages together. The housing market may also have an impact in preventing mobility of individuals from some small rural work-poor areas. However, the reasons appear to vary somewhat between work-poor areas. Therefore, any strategies that policy makers formulate to try and increase the hours of work of people in these areas may need to vary, taking into account their particular set of characteristics.

While increased educational attainment for individuals is likely to improve their life chances, the rapid rise in educational participation in tertiary institutions over the last decade appears to be unevenly spread on a geographic basis. This latter trend suggests that, increasingly, people with similar levels of education are clustering together. This has a major impact on some disadvantaged communities. For example, in these communities it will be difficult to draw together a Board of Trustees that includes people with the types of qualifications and expertise needed to oversee the running of a complex enterprise.

Finally, there is evidence that a significant proportion of both work-poor and education-poor areas remain in this state for relatively long periods in New Zealand. While there is mobility for some communities, it is generally fairly limited. However, it is not clear from this research whether the same people reside in these areas or, just as importantly, similar areas, for long periods. While some overseas studies suggest a high level of mobility for individuals, longitudinal studies as well as other types of research with a geographic dimension are needed to answer this question in New Zealand. Such research will help determine whether social policy should focus increasingly on trying to "strengthen" disadvantaged communities or primarily on endeavouring to improve educational and employment opportunities at an individual and/or family level.

(1) I would like to thank Statistics New Zealand for access to census data through my appointment as a Research Affiliate and Victoria University for providing financial support for undertaking this doctoral research. In addition, at various stages of the research Philip Morrison, Judith Davey, Gary Hawke and Kay Goodger provided helpful advice.

(2) This theory is still being debated as it is difficult to separate out the various effects of family norms, norms reinforced by schools and the wider neighbourhood effects (e.g Haveman and Wolfe 1995).

(3) The direction of causation is unclear, however.

(4) Morrison's research in the Wellington region has also shown that as individual Maori increase their level of education they tend to move to areas where people have similar educational qualifications, rather than staying in areas with a high proportion of Maori (Morrison 1993b).

(5) In choosing this measure, the relationship between the percentage of the prime-age population in each area not in paid work and the calculated average hours in paid work was explored. In 1986, a regression analysis had an [r.sup.2] of 0.58, but by 1996 the [r.sup.2] had increased to 0.80.

(6) For example, in Balfour Community in the South Island, there were 69 prime-age people in 1986 and 63 in 1996, and average hours increased from just under 30 in 1986 to just over 40 in 1996.

(7) In Auckland, as in some other larger urban areas, there tends to be a clustering of work-poor areas, with for example Glen Innes East, Point England and Tamaki neighbours. However, Glen Innes East is also a close neighbour to the high-income suburbs of St Heliers, Glendowie and Kohimarama.

(8) As it is not possible to divide 1,636 area units equally into deciles, the bottom three contained 164 units, the next four 163, and the top three 164 units.

(9) However, as shown by Figure 1, the gains and losses in average hours were not strongly concentrated in any particular decile. The percentage of area units in each decile which increased their total average hours between 1986 and 1996 were (from work-poor to work-rich deciles): 32%, 43%, 51%, 49%, 48%, 45%, 38%, 45%, 44% and 40%.

(10) This American longitudinal study also indicated that, between 1979 and 1985, despite some "churning" the poor urban areas were becoming poorer.

(11) Rural centres have a total population between 300 and 999. These are not termed as urban and identifying these settlements distinguishes them from true rural areas and larger townships.

(12) The hours of paid work measure includes unpaid work in a family business.

(13) It is debatable whether two separate job markets exist, one for men and one for women, or whether this trend simply reflects a general lack of labour demand in these areas.

(14) Wadestown is also the home of many of the architects and senior administrators of the reforms that took place over the period studied.

(15) However, if employment opportunities are plentiful it can be easier for single people either alone or with dependent children to financially support themselves and not need a partner for such support.

(16) In this analysis, both de facto and legally married couples are counted.

(17) For some of these couples being work-poor may not be a problem. For example, they may have chosen to retire early, and have a high level of savings.

(18) In a United States context, it has been suggested that when work-rich and work-poor areas are in close proximity they may have a high level of interdependency. Some groups of work-rich people have the money, and the need, to purchase domestic support services, while people in the work-poor area can potentially provide these services (Van Kempen and Marcuse 1997). However, this type of work is generally low paid, and often intermittent and part of an "informal" economy. As part of the informal economy it may not be recorded in official statistics such as the census.

(19) This measures unpaid work performed for people who did not live in the unpaid worker's household and includes a wide range of activities such as gardening for an elderly person or fundraising for a marae.

(20) Linked to this, as shown in Callister (1998), people with similar levels of qualifications tend to form couples.

(21) For women the [r.sup.2] was 0.35. This supports the The Economist's view cited at the beginning of the paper that it is the communities with large numbers of prime-age men out of paid work which are the most deprived.

REFERENCES

Brookings Institution (1997) Fortress America: Gated Communities in the United States, Brookings Institution, Washington, D.C.

Callister, P. (1998) "`Work-rich' and `Work-poor' Individuals and Families: Changes in the Distribution of Paid Work from 1986 to 1996" Social Policy Journal of New Zealand, Issue 10, pp.101-121.

Case, A.C. and L.F. Katz (1991) "The Company You Keep: The Effects of Family and Neighborhood on Disadvantaged Youths" Working paper No 3705, National Bureau of Economic Research, Cambridge, Massachusetts.

Castells, M. (1996) The Information Age: Economy, Society and Culture, Volume 1, Blackwell, Cambridge, Massachusetts.

Coleman, A. (1997) "Urban Joblessness, Location, and Hysteresis -- A Theoretical Approach" in P.S. Morrison (ed.) Labour, Employment and Work in New Zealand 1996, Proceedings of the Seventh Conference November 28 & 29, 1996, pp. 61-66, Victoria University of Wellington.

Crampton, P., C. Salmond and F. Sutton (1997) "NZDEP91: A New Index of Deprivation" Social Policy Journal of New Zealand, Issue 9, 186-193.

Cutler, D.M. and E.L. Glaeser, (1995) "Are Ghettos Good or Bad?" Working paper No. 5163, National Bureau of Economic Research, Cambridge, Massachusetts.

Department of Internal Affairs (1997) Building Strong Communities: A "Thinkpiece", Department of Internal Affairs, Wellington.

Department of Statistics (1992) New Zealand Standard Areas Classification 1992, Department of Statistics, Wellington.

Gramlich, E., D. Laren and N. Sealand (1992) "Moving Into and Out of Poor Urban Areas" Journal of Policy Analysis and Management, 11 (2):273-287.

Gregory, R.G. and B. Hunter (1995) "The Macro Economy and the Growth of Ghettos and Urban Poverty in Australia, Discussion Paper No.325, Centre for Economic Policy Research, Canberra.

Haveman, R. and B. Wolfe (1995) "The Determinants of Children's Attainments: A Review of Methods and Findings" Journal of Economic Literature, 33:1829-1878.

Herrnstein, R. J. and C. Murray (1996) The Bell Curve: Intelligence and Class Structure in American Life, Free Press, New York.

Mare, D. (1995) "Examining Labour Market Outcomes for Maori" Labour Market Bulletin, 1:116-123

Mills, E.S. and L.S Lubuele (1997) "Inner Cities" Journal of Economic Literature, 35:727-756.

Morrison, P.S. (1993a) "Supermap2: Scope and Potential with Special Reference to the Analysis of Income" Working paper No 21, Department of Geography, Victoria University of Wellington.

Morrison, P.S. (1993b) "Exploring Contemporary Gender and Ethnicity Issues: Making (Much Better) Use of Supermap2" Working paper No 22, Department of Geography, Victoria University of Wellington.

Morrison, P.S. (1995) "Work and Residence in Urban Labour Markets" in P.S. Morrison (ed.) Labour, Employment and Work in New Zealand: Proceedings of the Sixth Conference, 1994, pp.249-264, Victoria University of Wellington.

Morrison, P.S. (1997) "Unemployment and Non-employment: Urban Inequalities in New Zealand 1981-1991", paper presented to the Second Institute of Australian Geographers/New Zealand Geographical Society Joint Conference, University of Tasmania, Hobart, Australia.

Parker, B. (1997) "Very Long-term Job Seekers' Barriers to Employment: A Nationwide Survey" Labour Market Bulletin, 1:63-69.

Putnam, R.D. (1993) "The Prosperous Community: Social Capital and Public Life" The American Prospect, 13:35-42.

Reich, R.B. (1993) The Work of Nations: Preparing Ourselves for 21st-century Capitalism, Simon and Schuster, London.

The Economist (1997) "When Home is a Prison", 14 June, p. 92.

Van Kempen, E.T. (1997) "Poverty Pockets and Life Chances" American Behavioral Scientist, 41(3): 430-449.

Van Kempen, R. and Marcuse, P. (1997) "A New Spatial Order in Cities" American Behavioral Scientist, 41(3): 285-298.

Wilson, W.J. (1987) The Truly Disadvantaged, University of Chicago Press.

APPENDIX

Unit of Analysis

Area units must either define, or aggregate to define, territorial local authorities, urban areas and statistical areas. In some of these aggregations, a sense of "community" is sought with, for example, minor urban areas needing to provide five out of eight possible services such as a school or community centre (Department of Statistics 1992). In addition, area units of main or secondary urban areas generally coincide with suburbs or parts of suburbs which, for some people, will have some sense of "community", perhaps centred around a church or shopping centre. Finally, while the maximum total population of an area unit is about 5,000 people area units vary in physical size and population.

There were 1,775 area units in 1996, and the 1986 and 1991 censuses were recoded to the 1996 boundaries. Some of these area units either did not contain prime-age people or had very few people in them. Therefore, for this research, the smallest units were amalgamated to form a group classified as "other". The following criteria were used for placing small geographic areas in the "other" category. 1991 was used as a low point in overall employment. In this year, if there were less than 10 women or 10 men in the 25-59 age group who were gainfully employed in each area then the area unit was placed in an "other" group. The 1991 grouping was then applied to 1986 and 1996 data. However, in these years, if an area unit included in 1991 had less than six women or six men gainfully employed in the 25-59 age group, then, this area was also moved to the "other" category.

This left a total of 1,636 area units that were analysed. In this research, I also use data for usual residence rather than a person's census night location.

The Work Status of Area Units

One measure of employment in an area unit might be the percentage of prime-age people who are unemployed and seeking work. Using an "unemployed and seeking work" classification is problematic because this can be defined narrowly through setting strict criteria for determining how much a person has actively looked for a job. In addition, this measure will remove those who are not in the labour force who, in some communities, represent a significant group. For example, Morrison (1997) found that in the Wellington metropolitan area there was not only a widening gap in inequality between 1981 and 1991 when rates of unemployment were used, but the inequality increased when those not in the labour force were included in the analysis. However, there are also some problems of including people who are not in the labour force. For example, an area unit may contain a large educational institution where many prime-age people are studying, and who may, therefore, not be in the labour force. This type of work-poor geographic area will be quite different to one where people are not in the labour force because they are discouraged job seekers.

Paul Callister

Sociology and Social Policy Department Victoria University of Wellington