Maori in the Labour Market — 1991.

<>This is a draft report prepared in July 1994. The layout may have been corrupted; apologies.)


Keywords: Labour Studies; Maori;




This report is based upon the largest data base in New Zealand, the quinquennial population census. Te Puni Kokori requested from the Department of Statistics New Zealand a set of tabulations from the 1991 component of that data base, which would enable them to obtain a better understanding of the situation of the Maori in the labour force. Typically the tabulations compare the Maori (who for these purposes are any census respondents who describe their ethnicity as whole or part Maori) with the non-Maori. The tabulations were supplemented by some others from the Household Labour Force Survey for 1991 and 1993.


I was then commissioned to write a short report – actually a somewhat shorter than this one – on some of the features of the tabulations.


This presented a problem in that such is the richness of the data involved, it is not possible to report on every table. No doubt the tabulations will be used for many years to come, by researchers for a multitude of purposes. Any single report has to be selective. Such selecting requires some theme.


I decided at an early stage that it would be little point reporting that the Maori was an inferior position in the labour force in New Zealand. This fact has been known for many years and regularly reported upon. It would be a waste of the new data base to simply confirm an old story.


There was some attraction in attempting to assess the extent to which the Maori is “catching up” (or not) relative to the non-Maori. It is an important question, but for reasons discussed at the beginning of Chapter 4, it is not really possible to do this with a data base describing a single point of time (See also Chapter 2). To answer the question really requires comparable tabulations from  the 1986 population census (earlier ones do not have the same ethnicity question), and perhaps the 1996 one.


I report here that I did some preliminary investigations, and concluded that there is a sense in which it is true that the Maori is improving their employment performance, in terms of the characteristics of the jobs they hold, but so are the non-Maori. It is not evident the Maori is catching up the non-Maori, or if they are doing so it is doubtful that they are doing so rapidly. In my view a careful study to evaluate to what extent, and what ways the Maori is (or is not) moving towards a non-Maori labour force profile would be a valuable exercise well worth undertaking, if the more extensive data base was available. However whatever the conclusion, an uncontroversial result will surely be that the gap is still great, and that parity in whatever form that it may eventually take is some decades away.


This led me to consider what light the data base might throw on the processes which are generating the gap. I was very fortunate that Statistics New Zealand also supplied me with some econometric work carried out by one of their mathematical statisticians, Diane Craig. Its conclusions, which are reported in Chapter 3, are – I think – dramatic and new. Using the available personal characteristics, such as age, location, qualifications, gender, and family situation, Diane’s work demonstrates that only 30 percent of the difference in employment rates between the Maori and the non-Maori can be explained by differences in their personal characteristics. Of course the Population Census does not record all personal characteristics, but it is hard to think of a scientifically valid ones – except health, and perhaps psychological ones (but I am less qualified to comment on these) – which might make a significant difference by their omission.


The alternative is that the individual’s labour market status is not simply a result of the individual’s characteristics, but of the individual in a social context. This is the theme of Chapter 2. As I began to develop this thesis, I became aware that the conventional categorization of labour market status of employed, unemployed, and not-in-the-labour force is not necessarily the most helpful way to think about the issues which confront the Maori. First it is too static, but in addition it does not recognize that often the crucial difference is whether one has a good quality job or not. This distinction was compounded when I was making a routine check on labour force participation rates, and found that the Maori ones were much lower than the non-Maori ones. There is a sense in which there is a disguised form of unemployment which is large enough among the Maori relative to the non-Maori to seriously under-estimate the size of the Maori unemployment. Chapter 1 suggests that rather than unemployment rates of around 25 percent as conventionally measured, the Maori rates are nearer 40 percent.


It may seem odd for the Prologue to traverse the chapters in the reverse order to the text. In one sense it reminds us that theory construction and theory presentation rarely run in parallel. But it also arises because given the significance of the tabulations, I was keen to extend conventional theories of the Maori in the labour force – indeed the mechanics of the labour force for Maori and non-Maori – building on past knowledge and stimulating new ways of think about the issues.


Thus the Epilogue as yet unwritten tries to offer an new synthesis to the workings of the labour market as far as the workers in the market are concerned. In a way I am challenging the conventional wisdom, at least as presented by economists, but the real aim is of progressing our understanding developing out of the old one, using data that had not previously been so easily available.






According to the 1991 Population Census, the Maori male labour force was 87,519, of whom 66726, had either a full time or part time job and 20796 did not, being unemployed and actively seeking work. The ratio of unemployment to the total labour force (i.e. 20796/87519) gives a Maori male unemployment rate of .238 (23.8 percent). The rates for the Maori and Non-Maori by gender are given in the accompanying table 1.




Social Group               Labour  Force  Employed        Unemployed   Unemployment Rate

Maori Male                  87,519             66,726             20,796             0.238

Non-Maori Male         798,258           728,346           69,912             0.088

Maori Female              65,739             49,482 1          6,254               0.247

Non-Maori Female      612,654           555,846           56,808             0.093

Source: Table A2.


What the table shows is that the unemployment rate for the Maori compared to the Non-Maori is 2.7 times higher for males and females.


However before accepting this at face value, we need to ask about what is happening to those who are not in the labour force. This includes people who would take up a job if the opportunity arose, but who are not actively seeking work (perhaps because they have become too discouraged). A means of examining this is to look at the labour force participation ratios for the various groups. For example there are 131,742 Maori males over the age of 15, of whom 87,519 are in the labour force. The ratio (87519/131742) gives a labour force participation rate for Maori males of 66.4 percent.


This compares with a participation rate of 70.6 percent for non-Maori males. One difference between the two is explained by the differing age structures. The non-Maori are on average older, and so more likely to be retired. Suppose we apply the age-specific non-Maori participation rates to a labour force which had a age structure like the Maori. Then we would get a figure for the non-Maori of 76.6 percent, higher because the Maori has relatively more in the peak working ages. The results of applying this analysis to all the four gender-ethnic mixes is shown in Table 2.




Social Group               Labour  Force  Population*     Participation Rate       Participation Rate

(Maori Age                  (N-M Age

Structure)                   Structure)

Maori Male                  87,519             131,742           0.664                           0.766**

Non-Maori Male         798,258           1,130,346        0.605**                       0.706

Maori Female              65,739             140,061           0.353                           0.603**

Non-Maori Female      612,654           1,188,138        0.405**                       0.515

Source: Table A2. * Over 15. ** indicated estimated on other population age structure.


The figures show that actual participation rates for the Maori are lower that for the Non-Maori, but if they had had the same age specific rates they would be higher, because a higher proportion of Maori are in the peak working age groups.



Using Non-Maori Age Specific Participation Rates

Social Group               Labour Force   Employed Unemployed**      Adjusted

Unemployment Rate

Maori Male      100,941*         66,726 34,224*           .339*

Non-Maori Male         798,258           728,346           69,912             .088

Maori Female  84,478*           49,482 34,987*           .414*

Non-Maori Female      612,654           555,846           56,808             .093

Source: Table A2. * indicated estimated on a non-Maori age specific labour force participation rates. ** and not reporting


Some confirmation for the hypothesis that the Maori have more discouraged unemployed than the non-Maori comes from the Household Labour Force Survey estimates of the jobless who are not definitionally unemployed. In both years for which the data is supplied, the Maori proportion of the working age population was higher than the non-Maori proportion: 8.1 percent for the Maori against 6.7 percent for the non-Maori in March 1991; and 13.9 percent against 7.0 percent in March 1993.


Suppose we assume that the Maori actually had the same age specific participation labour force participation rates as the Non-Maori, in effect treating the depression in the participation rates as there are Maori who were unemployed but did not report their state as such. Adding them to the total unemployment, we obtain unemployment rates for the Maori of 33.9 percent for males, and 41.4 percent for females. If the Maori had the age-specific labour force participation rates of the non-Maori, their unemployment rate would be 3.9 and 4.5 times the non-Maori rates respectively.


We should be cautious in using such figures. They do not conform to official definitions. They may reflect other effects such as the Maori more likely to live in areas where there are so few jobs that everyone there – Maori and non-Maori – are more likely to be discouraged from seeking work rather than being unemployed in the technical sense. It is also to be emphasized, that the figures at best simply reduce the Maori rate of discouragement to the non-Maori rate. There are others in each group who are also discouraged from seeking employment by their labour market situation.


Nonetheless the figures suggest that Maori unemployment is proportionally higher than the raw data implies. Alternatively it suggests we cannot just look at the labour force, but need to take into consideration those not-in-the-labour-force, when trying to understand unemployment.


Appendix: Labour Force Participation Rates


This appendix provides graphs of the labour force participation rates by ethnicity and gender.


Figure 1: Male Labour Force Participation Rates


Figure 1 shows the results for males in five year age cohorts. In each cohort the Maori rate is below the non-Maori rate. Up to the age of 50 they are between 10 percent and 14 percent below. Thereafter the percentage rises, perhaps because of earlier Maori retirement – for health or other reasons. In the 61 to 65 age cohort the Maori participation rate is 30 percent below the non-Maori one.


Figure 2: Female Labour Force Participation Rates


Figure 2 shows the results for females. Not unexpectedly the participation rates for both groups are lower than for males. There is also the characteristic M formation for each, as women reduce their involvement in the paid labour force during peak child bearing and rearing years. However the dip at the middle of the M is less than in the past.


The Maori dip is in the 26 to 30 age cohort, and the non-Maori is in the 31 to 35 cohort. As a result the female Maori participation rate is not as uniformly below the non-Maori one in the peak working years, varying between 83 percent (in the 36 to 40 cohort) and 72 percent in the 26 to 30 cohort. In the later years there is not the same pattern as for males, with the ratio actually rising as non-Maori women retire earlier than their menfolk. In fact Maori women over the age of 76 are more likely to work than non-Maori although the numbers are so small the rates are not shown in the graph.








The tabulations which accompany this report come from the largest and most comprehensive data base published on the Maori in the labour market. It could be, and for various research purposes will be, supplemented by special surveys, and data from earlier Population Censuses and Household Labour Surveys. But by itself there is no larger data base, nor one whose definitions are as systematic and rigorous.


Yet they provide only a snapshot, or at best three snapshots, of the labour market. Because we have a habit of thinking that the pictures are the reality, it is useful to explore an extended metaphor to understand the snapshot’s weaknesses and strengths.


Suppose a class was told to jump up and down on the floor, and the teacher took a photograph to analyze the student behaviour. The picture would show each child either on the floor or off it. Suppose our interest was gender differences, and the picture showed the boys were less likely to be on the floor than the girls. We might worry this was an accident due to the picture being taken at a particular moment in time, but if the numbers were sufficiently large conclude that the difference between the genders was statistically significant.


Now the snapshot does not capture the dynamics of the jumping. We would not be able to tell from the photo, for instance, if the class was jumping to music and so there was a degree of coordination between the jumpers. Nor could we find out whether vigorous jumpers jumped for a shorter period. Note too, that by asking the question whether the student was off or on the floor we overlook the height of the jump. Perhaps those who jump highest spend least time on the floor, or have a longer retirement at the end. The extent that the single photo adds to our understanding depends on the importance of the dynamics to the situation. The danger is the single shot approach entraps us into asking only limited static questions of questions, and we ignore the essential dynamics. The same is true for the labour market. The information we have is mainly of the snapshot variety, but that does not mean the dynamics are insignificant or should be neglected.


This report is written in the belief that labour market dynamics are crucial, but that the snapshot of the single instance data base can give us some useful insights, especially if they can be supplemented with other (more longitudinal) data.


While we have the metaphor there is another critical issue to be illustrated. A “fact” is open to a number of interpretations. The obvious one given the difference between the boys’ and girls’ jumping is that boys are more athletic and jump more vigorously. Even were that true, we could not tell from the evidence here whether there was some genetic predisposition, or whether boys were encouraged to behave that way and girls were not. But there are other explanations. For instance the teacher might report that the girls tend to resent the boisterousness of the boys, so that the girls jump higher and more often when they are by themselves. The teacher might report the boys, on the other hand, jump excessively to show off to the girls: by themselves they are much less vigorous. Or suppose we had another photo with the girls more off the floor, when the teacher told them that time to dance with graceful leaps. This time the boys objected to the instruction as cissy, while the girls took it up with enthusiasm.


The point of the last paragraph is it illustrates that a fact itself tells us something, but typically we interpret it in a context with a number of explicit or implicit hypotheses. The serious social statistician has an obligation to be as explicit as possible, and to at least mention alternative hypotheses which may be valid, as best as he or she can. This may be irritating to some readers who simply want the “facts”, but they can acquire them directly from the tabulations. Even so they need to be aware of the implicit hypotheses they use when interpreting their facts.


The Statics of the Labour Force


What does the snapshot from a survey of the labour market show? In broadest terms each respondent is categorised into one of three categories:

– the employed are those who are gainfully employed in the full-time or part-time labour force;

– the unemployed are those who are not employed but are actively seeking work;

– the not-in-the-labour-force (NITLF) are those who are neither employed or unemployed;

which involves the subsidiary definition

– the labour force are those who are employed or unemployed.


These are international definitions, with the advantage that they facilitate international comparisons and are less vulnerable to political manipulation. There are detailed, and internationally agreed criteria, for dealing with particular circumstances.


But the definitions can generate paradoxes to the unwary. A person on the unemployment benefit may be “unemployed”, but they may be “employed” if they are working an hour or two a week (as the benefit allows them), or “not-in-the-labour-force” if they are no longer actively seeking work.  This is an illustration that a definition for administrative purposes may not reflect a rigorous scientific framework.


One response to this is to create ad hoc definitions for particular purposes. Statistics New Zealand uses a notion of

– the jobless which adds to the unemployed those who were available for work but not actively seeking it (perhaps because they were discouraged), and those actively seeking work but not (immediately) available for work (perhaps they are in an educational institution).


Other definitions are available (e.g. what about those who are employed but would like to work more hours?). We will return to the jobless in a later section, but most of the report, and the data, focuses on the three way split of the population.


What do the statistics mean? At the simplest they report the number in each category. It is also common to report the

– unemployment rate, the percentage of those in the labour force who are unemployed.

Those NITLF do not appear in this measure. This means that if some of the unemployed get so discouraged they give up actively seeking work the unemployment rate may go up.


Ignoring the treatment of the NITLF the unemployment rate being the percentage of the unemployed measures its prevalence. For two different populations – say Maori and non-Maori – it tells us how much proportionally greater the prevalence is in one group compared to another. However, and this is crucial, differences in the unemployment rates between groups give no indication of the differences in the quality of unemployment.


This is a simple – indeed obvious –  proposition which has tended to be overlooked as we became mesmerized with the numbers of those in the box labelled “unemployed”.


It is possible, of course, that the average quality of unemployment does not vary with the level of unemployment. But even were this true – we have little evidence to accept or reject the proposition – it seems unlikely that the average quality is the same for different social groups. To take a non-controversial case, it seems likely that in the 1950s and 1960s unemployment was not as harsh a condition for women as it was for men (or is today).  It seems likely that the quality of unemployment is harsher today than it is for Maori, something which will be deduced as the data is reported.


The Dynamics of the Labour Force


Thus far there has not been a definition of the quality of unemployment. It is best provided in the context of a dynamic interpretation. The introduction of the notion of the quality of unemployment is not to try to make the state of unemployment appear more onerous than it is. Undoubtedly for many of the unemployed the situation is extremely onerous. There is good evidence to show that unemployment correlates with morbidity and mortality (especially suicide) both for the unemployed and the unemployed’s family.  However that is not true for all unemployed, and it was probably not true in the 1950s and 1960s.


Then the unemployed had a reasonable chance of obtaining work not long after they required it. Certainly there might be the trauma of redundancy, and some anxiety about obtaining the next job (plus perhaps a little more about whether they would like it), but such psychological pressures were much less than today for the averaged unemployed.


Yet today there are still workers who become unemployed and yet do not suffer severe trauma. Typically they are well skilled with prospects of a good job if they search conscientiously, and with adequate finances (perhaps out of their savings and redundancy payments or with the support of a spouse). They may even enjoy the forced holiday as they diligently and confidently seek alternative employment.


That this occurs for some but not others has profound implications for the meaning of a full employment economy. In the future, as in recent years, there will be redundancy as firms respond to technological change, demand shifts, and external market movements. Some of the redundant will be temporarily – unemployed. In a dynamic growing economy full employment is literally impossible, since there will always be some workers without a job. Yet we may be well satisfied if the unemployment is high quality in the sense that the unemployed and their families are not under psychological or economic stress, and if their morbidity and mortality are not elevated as a result.


That ideal stated, there are the unemployed who are under severe stress today. We would like to divide out according to some measure of quality of their state of their unemployment, but that is not easy with the current data bases. One attempt is to separate out

– the long term unemployed, usually defined as those who have been unemployed for more than six months.

However there are at least two major omissions with this definition, both of which involve a more dynamic labour force than casual view at the snapshot implies.


First there are the unemployed who have not yet been unemployed for six months, but have the realistic expectation that ultimately they will be. They must be under similar stress to the long term unemployed. Second there are those who experience frequent bouts of unemployment, even though they may never have been unemployed for, say, six months at one time. A person who has experienced a total of six months or more unemployment in the last two years – say – may be as stressed as someone who is unemployed once for the same period.


It is not proposed here to deal further with the unemployed in the first category.  The second category raises an even more substantive point.


Not only are there different levels of quality in the unemployment experience, but there are different levels in quality of the employment experience. The conventional way economists describe this phenomenon is to divide the employed labour force into two, called the primary and secondary markets, adding the unemployed to the secondary market, which consists of the low quality jobs and the unemployed.  Those in the

– primary labour market have good pay, working conditions (including job security), and good career prospects.

On the other hand those in the

– secondary labour market experience poor pay, poor working conditions, poor career prospects, job insecurity, and frequent unemployment.

This is distinction is the underlying notion in the theory of the dual labour market.


The characteristic experience of those in the secondary market is a continual churning from employment to unemployment and back again. Being employed may be better than being unemployed, but not that much better. In any case they may soon be unemployed again.


Further detail of the theory appears in an appendix. The point to be made here is that not only does the Maori appear relatively more frequently among the unemployed, but it seems likely that they are also relatively more involved in the poor quality jobs in the secondary labour market, and that they experience more labour market churning. Of course this hypothesis does not deny that there are non-Maori, including Pakeha, who are also in the secondary labour market.


The existence of a secondary labour market has policy, as well as empirical, implications. We might ask to what extent is employment policy merely about shifting the unemployed into employment the secondary market, speeding up the churning rather than offering the unemployed prospects in the primary labour market. This may be better than doing nothing, but is it the best we can do?


Or to put the same point more positively, we might ask can employment policy be reoriented to enable those who go into employment in the secondary market, the opportunity to shift themselves into the primary market, thus breaking out of the loop of churning from unemployment to employment and back.


The Theory of the Dual Labour Market


Dual labour market theory argues the labour market can be segmented into (at least) two parts, flows between which are seriously impeded.  It is not difficult to see a job quality spectrum in the work force with at one end a set of `good’ jobs and at the other end of `bad’ jobs. At the top end working conditions are characterised by high pay, promotion prospects, good security, quality working conditions, and attractive fringe benefits. At the other end the workers have jobs which are poorly paid, with little prospect of advancement, poor working conditions, and insecure tenure. Such work is often intermittent and its workers can be frequently unemployed.


At issue is what happens in the middle of this spectrum. As Robert Bowie has pointed out the distribution of the jobs along the quality spectrum could be uni-modal, suggesting a considerable connection within the work force, or it could be bi-modal suggesting a segmentation of the labour force into primary and dual labour markets.  (Figure 1) Which is in practice most true is an empirical question which cannot readily be resolved.


An alternative approach is to investigate the extent to which the labour force is interconnected. It is idle to assume that an unemployed teenager in Kaitaia is a ready substitute for a senior Treasury official in Wellington, but we might ask how the change in either’s work situation influences the work situation of the other. Any answer involves a time dimension. We would not expect an immediate response at the lower quality end of the labour market to a change at the higher quality end, but we could envisage that the changes percolate through over time. But that raises the empirical question as how far they do, and how long it takes.


If changes transmit quickly, the labour market may be treated as reasonably homogeneous – as is implicitly assumed in much economic discourse. If the time involved is years or even generations, then practically the labour market may be treated as heterogeneous and segmented.


A worker’s place on the quality spectrum can change over time. Many adolescents in low quality jobs in the retail and fast food firms are tertiary students in training for quality professional positions. But it is also possible that work experiences and on the job training enable workers to upgrade their position, moving steadily up the work quality ladder. In the following I shall for simplification ignore the life cycle patterns – most evident in the adolescent story – but include the upgrading process within the labour force. I shall also ignore the complications of the internal labour market, and the peripheral work force.


What systematic evidence do we have of the degree of segmentation in the New Zealand labour market? Using existing statistics Bowie provides “circumstantial evidence” that there is a group whose working conditions were analogous to those in a secondary labour market. He noted that women and ethnic minorities appear more likely to be among that group.


At about the same time sociologist Susan Shipley was surveying 750 households in Palmerston North, mainly from the perspective of women in the labour force.  She found that part time workers were more predominantly female and that gender segmentation was more marked in the full-time labour force, and argued that this pointed to labour market duality. There is a tendency in her work,  to equate part-time work with the dual labour market. It is of course only part of the totality of the secondary market, and some part-timers – such as professionals with a high leisure preferences (or phasing into retirement) – may be working part-time in the primary market.


The evidence that women and men have different occupational profiles and that it was taking such a long time to break those differences down, suggests that there was not a great deal of mobility in the labour market, except slowly through time.  A dual labour market would be the minimum segmentation required to explain this gender differentiation.


More recently two statisticians at the Department of Statistics used income tax data to identify income profiles over time.  They found that the 81.6 per cent of the salary and wage earners who were in the highest income quintile in the 1980 tax year were still in the top quintile in 1981. Seven years later 55.9 per cent were still in the top quintile. After various adjustments that suggests a relegation of top-income employees into lower income brackets of less than 3 per cent a year.


The picture for those lower in the income stakes is of greater instability, but there is no great expectation for advancement. Suppose an employee was in the middle quintile in 1980. Then the chances of being in the top quintile seven years later was 9.5 per cent. That probability includes young professional workers at university, or starting out on a career and quickly increasing their income.


The income statistics, which include the unemployment benefit, do not capture well the churning between employment and unemployment. A report prepared by the Department of Labour to the Prime Ministerial Task Force on Employment provides some insights, using  the data from the registered unemployed.  Over the period from October 1988 to June 1993, 754,312 enroled on the New Zealand Employment Service register. To give some idea of this magnitude the average size of the labour force was about 1,612,000 people, so this number represents about 47 percent of that total, or around 10 percent of the labour force each year.


Those that enroled more than once, are counted but once in the above total. In fact over 45 percent were enroled at least twice, and 2.1 percent more than 5 times. This suggests that at least 21 percent of the labour force experienced repeated unemployment in less than five years period, and 1 percent experienced it on five or more occasions. The average number of enrolments was 1.78 times for non-Maori, and 2.18 times for the Maori. While the average cumulative duration on the register was 59.2 weeks, the Maori averaged 68.8. weeks. The register does not pick up all the churning, but the available data is indicative that it was substantial in recent years.


Altogether then we have tantalising evidence which is not inconsistent with the dual labour market hypothesis.

The Direction of the Study


The implication of all this is that we are not interested in the snapshot of the labour market in itself, but in what it can tell us about the dynamics of the labour market. Because we cannot trace the dynamics directly we are going to rely on the theory of the dual labour market to interpret the snapshot. Thus we are not going to be interested merely in whether a person is employed or unemployed, but what is the quality of their experience in those states. Since we cannot assess the quality directly we will have to use indirect measures: in particular that certain occupations and other personal characteristics are associated with superior or inferior quality jobs.





If we define the employment rate as the proportion of a population (or subgroup) who have employment, then the Maori has a low employment rate. We use this definition to avoid the impact of variations in the labour force participation rate between subgroups.


One of the reasons the Maori experiences a lower employment rate than the non-Maori is because their situation is worse off, in regard to a number of key personal characteristics important to acquiring work. Proportionally more Maori are among the young, who are less attractive to employers; fewer have high educational qualifications which are associated with better job prospects; more live in high unemployment areas.


Thus low employment may not be merely associated with Maoriness per se, but because the Maori is more likely to have characteristics which are associated with not being employed (either unemployed or not-in-the-labour-force). Insofar as this is true there is a tendency to overstate the severity of Maori unemployment, by confusing it with the severity of unemployment among the youth, the unskilled, and those in the most marginal regions, and so on.


Standard statistical techniques enable us to separate out the effect of these other variables from the effect of Maoriness on the likelihood of being employed. The econometric description is in an appendix, but the next few paragraphs provide a layperson’s guide to the underlying ideas.


An employment rate might be thought of as a probability. Suppose we take at random one person in the 15 to 24 age group. According to the 1991 Population census the probability was 48.0 percent he or she would be in employment. The employment rate for the age group was 48.0 percent. The exercise can be applied to any subgroup. The probability (employment rate) for a Maori in the age range was 32.4 percent, and for a non-Maori was 51.2 percent.


In principle we can do this for very detailed subgroups, providing we have the data. For instance we might consider the probability of being employed for a 15 year old female non-Maori, who lives in Auckland City, has no educational qualifications, and is a dependent child. Or what about a 24 year old Maori who lives in Nelson, has a university degree, and is a partnered father? The total number of combinations is large.


Instead of tabulating each of the combinations they are summarised in an equation which adds the various effects together. It is assumed the effects are additive, and do not interact. That one has a degree, and are Maori are treated as independent of one another as they impact on employment. The equation starts of with a base situation, as follows. Consider a


15 years old;

living in Auckland Region;

in a city location;

as a dependent child;

with no educational qualifications; who is



The equation estimates that her expected employment rate would be 13.22 percent (round it to 13.2 percent).


As any of these characteristics change we add the difference that is tabulated in the appendix. For example suppose the person is

male rather than female: add 3.95 percentage points from the total (i.e. a male is more likely to be employed than a female all other characteristics the same);

24 years old, rather than 15 years old: add 27.04 percentage points (the probability of employment rises with age);

lives in the Nelson region rather than the Auckland one: add 5.33 percentage points (Nelson is the region with the highest employment rate, Northland is the one with the lowest);

lives in a rural location rather than a city: add 2.93 percentage points (for employment rates are is higher in the countryside);

is a partnered father rather than a dependent child: add 21.94 percentage points (not surprisingly);

has a university degree rather than unqualified: add 9.05 percentage points (as  a rule the more education, the more likely one is to be employed ); and who is

Maori rather than non-Maori: subtract 13.11 percentage points (as discussed below);

This person then would have an employment rate (i.e. a probability of being employed) of (13.22+3.95+27.04+5.33+2.93+21.94+9.05-13.11=) 65.02 or 65.0 percent.


The effects of the characteristics are much as we might expect.


What about the 13.11 percentage point difference between the Maori and the non-Maori? It tells us that if we have two people who are otherwise identical in terms of their characteristics reported above, the probability that the Maori is employed is 13.11 percentage points lower than for the non-Maori. To put this in perspective, earlier we noted the non-Maori employment rate for 15 to 24 year olds was 51.2 percent. If the Maori had had the same average characteristics as the non-Maori their employment rate would have been 51.2-13.1 = 38.1 percent. Since the actual Maori rate was 32.4 percent, only 30 percent  of the difference between the Maori and non-Maori employment rates can be explained by the Maori having different (and inferior) census characteristics compared to the non-Maori.


In a way this is quite an astonishing conclusion. It seems to say that Maoriness is a handicap to getting a job.  Before jumping to this conclusion we need to consider whether there is some design effect in the econometrics which invalidates the conclusion. They are not obvious.


Another set of explanations is that key variables not directly related to Maoriness have been omitted.  These omitted variables may not necessarily be personal characteristic variables.


A crucial issue in obtaining work may be the process by which it is obtained.  For instance there is some evidence that very often a critical step in acquiring employment is a network of people typically employed themselves who are informed about the local (or relevant) labour market. Thus if one’s family (or whanau) and social groups are unemployed, have poor quality positions , or are not-in-the-labour-force the individual has less information about job prospects and likely to remain unemployed. Insofar as this vicious circle is important, that the Maori as a group has lower rates of employment then each individual Maori dependent upon a Maori network has lower chances of obtaining work in comparison to a non-Maori in a better connected social network.


This network theory of obtaining employment is a scientific hypothesis insofar as one can envision systematic empirical investigations which may reject or support the hypothesis.  However there is little research into this or, in general, theories that the critical role of the process of finding work.  At this stage the investigations would not usually be so much statistical studies based on populations or large samples (as the one reported here). In the first instance the research is likely to be anthropological type studies of small groups of individuals, tracing their job seeking experience, and trying to determine the features which distinguish success from failure.


What the research reported here tends to suggest is that personal status variables alone are insufficient to explain the employment differences between Maori and non-Maori. If this is true, then a strategy of focusing on improving personal characteristics, through better education by itself may not be as successful as might be hoped. Other strategies may be necessary to tackle the higher level of Maori unemployment.


Or to put the research in another way. It would appear that differences in (available) personal characteristic variables explain only 30 percent of the difference between Maori and non-Maori employment rates for those in the 15 to 24 age range. The remaining 70 percent is yet to be explained, probably as much by social as personal characteristics.







In this chapter we compare the characteristics of the employed Maori (M) and the Non-Maori (N-M) in regard to their occupation, industry, training, and employment status.  The conclusion is not surprising. The Maori tends to be in the occupations with the highest unemployment (and relatively low expansion), in the slower expanding industries, and to be less well trained. In summary as a rule the Maori is in an inferior position in the labour force. In dual labour market theory they are located in the secondary labour market.


Inferring Longitudinal Behaviour From Cross-Sectional Evidence


As an earlier chapter emphasizes, the data we have is a snapshot in time. Thus the person born in 1970 is recorded at the age of 21 in the 1991 Population Census, but the person born in 1955 is 36. Thus any comparison between the cohort of 21 years olds and 36 year olds involves looking at two distinct processes – one cohort is 15 years older than the other, and the same cohort was born 15 years earlier. Since birth-date and age both affect behaviour, we cannot easily separate them out from a single observation.


Figure 3


Figure 1 illustrates this complication. On the vertical axis is some behaviourial measure – suppose it is income. The upward sloping lines show the experience of four cohorts, at the ages of 21, 26, 31, and 36. The lowest line refers to the cohort born longest ago, in 1955. Each successive cohort experiences a higher income when they begin, but the same upward slope.


The downward sloping line is what the snapshot of the population census in 1991 would show. Despite each cohort’s income rising through time it also shows a falling income between cohorts at any point in time. This is a paradox only insofar as we think cross-sections of individuals at different life stages can be directly interpreted to show the longitudinal trajectories of individual lives.


Fortunately we avoid some of these difficulties by comparing Maori with non-Maori. The totals are also given, but only used where they summarize the individual cohort differences. On occasions to interpret the patterns we have checked back to the 1986 Census, where that is possible, for there is not always comparable data.




In the census  those employed are classified nine major occupational groups, plus “not adequately defined” (NAD). Respondents are asked with regard to their main job “[w]hat is your occupation?”, and “[w]hat tasks or duties do you spend the most time on?”. These are then coded according to a standard set of classification rules, and then aggregated into the ten groups.

Table 1




AGE                15-24               25-44               45-59               TOTAL*

M         N-M    M         N-M    M         N-M    M         N-M

Managers &     1.9       3.2       6.7       13.4     7.4       15.9     5.6       12.1


Professionals   3.1       5.5       7.7       14.4     10.8     13.4     7.2       12.4

Technicians     6.3       9.7       8.2       12.3     6.3       9.8       7.3       11.1

Clerks              16.1     19.4     11.5     13.6     8.3       13.7     12.0     14.6

Service &        19.6     21.9     12.2     10.8     11.0     10.4     13.8     12.6

Sales Workers**

Agriculture &  8.6       7.7       7.9       9.3       7.8       10.7     8.2       10.0

Fisheries Workers

Trades Workers10.6    13.6     9.2       10.7     6.6       9.7       9.0       10.8

Operators        14.0     7.9       20.4     8.8       22.4     8.8       19.0     8.5

& Assemblers

Elementary      16.0     9.5       13.6     5.5       15.9     6.2       14.7     6.4


N.A.D.            3.7       1.6       2.7       1.3       3.6       1.3       3.2       1.5


Notes:  M = Maori,                              N-M = Non-Maori

* including 60+.                      ** including Armed Services

bold indicates larger in age group.

Source: Table A13.


Table 1 shows the proportion of each age group in each major occupation. The Maori are concentrated in the three least skilled occupations – Operator & Assemblers, Elementary Occupations, and NAD – where they are more than twice as common relative to their prevalence in the population. They are slightly more common among sales and services workers (including the armed forces). On the other hand they are less than two thirds as common among the top three classes – managers, professionals, and technicians – and also relatively less common among clerical workers, agricultural and fisheries workers, and tradespersons.


There are a couple of anomalies from this pattern.  Despite the relative Maori dominance in the sales and service sectors, they are under-represented here in the 15 to 24 year old group. Conversely the Maori in the same age group are over-represented among agricultural and fisheries workers.


This is almost certainly transition behaviour, also evident in the 1986 census. A possible explanation is that unskilled young Maori become farm employees after leaving school, temporarily boosting their sector proportion, until they are overtaken by non-Maori farm owners with more education. Meanwhile the reverse happens in the service and sales area, which young non-Maori school leavers go into, but apparently get (relatively) overtaken by Maori later. It is possible that proportionally more of the young non-Maori are there only temporarily, perhaps doing relatively low skilled work financing themselves through tertiary training, and exiting the occupation when obtaining their qualification.


One other curiosum deserves mentioning. The relativity between the Maori and the non-Maori of the managerial class is much the same after 25, at (roughly) half. However the relative proportion for profession rises with aging cohort. The apparent implications are that older cohorts of Maori have been more successful at becoming professionals than the current one or that Maori become professionals at a somewhat later age than non-Maori. Neither pattern is consistent with anecdote.


The figures are of course being affected by the faster falling labour participation rate of the Maori for older group, with the likelihood that the higher paid professionals hang on. But if that argument has any force, we must conclude that it applies also for managers, which means that the participation adjusted proportion for Maori managers is rising with the older cohort (or falling with younger ones. The puzzle does not seem resolvable with this data base, probably requiring a much finer categorization of managers and professionals.


The difficulty the average Maori faces in obtaining employment is illustrated by a comparison of the demand for jobs in those occupations where they are over-represented. According to the March 1991 Household Labour Force Survey, the service and sales workers, operators, and assemblers, elementary occupations, and not specified (NAD) made up 55.0 percent of the unemployed, whereas they were only 30.8 percent of the employed according to the 1991 Census. The Maori were even worse placed, with 65.4 percent of their unemployed chasing the 30.8 percent of jobs.


This report is reluctant to pursue policy conclusions in too greater depth, not only because that is not in the commission, but at this stage the research needs to aim at what is the “is”, not what are the possibilities. Nevertheless this section provokes the question as to the extent to which training for low level jobs, which are the sort of options commonly offered to the unemployed, are going to address the concentration of the Maori in low skilled occupations.




The industry share table contradicts the conventional wisdom in at least some respects. The image of the Maori is of a rural dweller in primary industries. (Table 2)


Table 2 INDUSTRY SHARES BY AGE Percent of Age Cohort


AGE                15-24               25-44               45-59               TOTAL*

M         N-M    M         N-M    M         N-M    M         N-M

Agriculture,     9.2       7.7       8.6       9.6       8.6       11.0     9.0       10.3

Hunting, Fishing, & Forestry

Mining .3         .2         .6         .3         .8         .3         .5         .3

& Quarrying

Manufacturing22.0     16.5     23.1     15.7     21.8     15.7     22.4     16.0

Electricity,       .6         .5         1.3       .9         1.5       .9         1.1       .8

Gas & Water

Building          5.6       5.6       7.1       6.3       6.5       6.0       6.6       6.0

& Construction

Wholesale        23.1     30.0     13.2     18.7     11.3     18.9     15.3     20.8

Retail Trade & Restaurants & Hotels

Transport,        5.0       4.8       8.6       6.4       9.6       6.1       7.8       5.9

Storage & Communications

Business          7.3       13.1     6.3       12.5     4.5       10.4     6.2       12.0

& Financial Services

Community,    22.7     19.8     27.9     27.4     31.7     28.9     27.4     26.2

Social & Personal Services

NAD               4.3       1.7       3.4       1.7       3.6       1.8       3.7       1.8


Notes:  M = Maori,                              N-M = Non-Maori.

* including 60+

bold indicates larger in age group

Source: Table A12


In fact there are relatively fewer Maori in Agriculture, Hunting, Fishing, & Forestry.  This conclusion is consistent with the occupational data of the previous section. The data is not detailed enough to be able to precisely state why this circumstance. A possible explanation is that the conclusion is dominated by farming, where the self employed and employers are more likely to be non-Maori. We conjecture that among the employees in the primary sector the Maori relatively predominate.


More generally there are relatively more Maori in all industrial sectors excepting primary, domestic trade, and business and financial services.


On the basis of the tabulation it might be concluded that the Maori should make greater efforts to penetrate domestic trade and business and financial services. However the industry imbalance is not as grotesque as the occupational one. It is not clear that the Maori will be markedly better off by switching into low level occupations in these two sectors, rather than seeking high level occupations in any sector.


Education and Training


Educational and vocational attainment is asked for the entire population, and not just for the employed as in the case for occupation and industry. In this section we use all the available information, covering the entire population. The recorded information is on their highest educational attainment, so a person with a post graduate degree might well have a number of other qualifications on the list.  (Table 3)


Table 3 HIGHEST QUALIFICATION (Entire Population)

Percent of Age Cohort


AGE                15-24               25-44               45-59               TOTAL*

M         N-M    M         N-M    M         N-M    M         N-M

Postgraduate   .2         .5         .6         3.4       .8         3.0       .4         2.0

Bachelors        .8         3.5       1.5       6.9       .9         3.0       1.0       4.2


Under Graduate.4       .9         .9         2.2       1.0       2.0       .6         1.6

Technicians     1.3       3.0       2.4       5.6       1.6       3.9       1.7       3.8


Teaching          .7         1.2       3.1       6.7       5.2       7.5       2.1       4.8

or Nurses  Certificate/Diploma

Trade               5.6       7.2       9.5       12.9     6.5       11.1     6.9       9.7


Other               9.4       9.8       8.2       9.9       5.5       9.0       8.3       9.2

Tertiary Certificate

Higher School 4.3       11.2     1.4       2.6       1.5       1.8       2.9       5.4

or Leaving Certificate/Bursary

University       10.2     16.7     5.1       7.9       1.5       2.7       7.1       9.1

Entrance/ Sixth Form Certificate

School 17.9     19.6     12.3     13.4     7.6       9.5       14.3     13.9


Other School   .8         1.2       .5         1.9       1.4       3.6       1.0       4.0


Still at School  8.6       7.8                                                       4.4       2.6

No                   36.9     15.4     53.2     25.4     64.2     41.1     47.0     27.8


Not Specified  3.0       2.0       1.2       1.3       2.5       1.6       2.4       2.0


Notes:  M = Maori,                  N-M = Non Maori.

* including 60+

bold indicates larger in age group

Source: Table A10.


The overall conclusion about educational and vocational attainment is not surprising. In every age group the non-Maori is markedly better qualified than the Maori. Because attainment of qualification continues throughout the individual’s life we must be careful in making comparisons between cohorts. Nevertheless the gap is so large it would seem that the Maori born between 1947 and 1966 are less qualified than a non-Maori born in the preceding twenty years between 1927 and 1946. An aphorism might be that the non-Maori of the previous generation is better qualified than the Maori of this one.


One might wonder whether the gap between those born after 1966 is as great. The data is complicated by many in this group are still obtaining qualifications. However a comparison between the gap between those in the 20 to 24 year old group, compared to those in the 25 to 44 year old group, does not suggest the younger Maori were relatively better qualified, except at the lowest level of attainment. There is now almost no relative gap between those whose highest qualification is school certificate, and the gap has diminished between those with only an “other school qualification”. This is probably due to the change whereby today it is possible to get a school certificate in a single subject. The true change may not be as great. In any case there is still proportionally more than twice as many Maori in the 20 to 24 age group without any qualification.


In simple terms the efforts the Maori has been making to improve their qualification attainment has been paralleled by that of the non-Maori. There is no obvious evidence from this data that the Maori are catching up to the average of their cohort.


Table 4 EMPLOYMENT STATUS   Percent of Age Cohort


AGE                            15-24   25-44   45-59   TOTAL*

M         N-M    M         N-M    M         N-M    M         N-M

Wages or Salary          96.3     95.9     88.2     77.6     86.0     71.3     89.6     78.2

Employer                     1.0       .8         3.7       8.8       3.9       11.6     3.1       8.3

Self-Employed            1.4       2.6       6.8       12.3     8.7       15.5     5.8       12.0

Unpaid Relative          1.4       .7         1.3       1.3       1.4       1.6       1.5       1.5


Notes: * including 60+


Source: Table B2.


The employment status data is not surprising insofar as it shows the Maori less likely to be employed or their own boss.  It may seem surprising that both groups have the same proportion of “unpaid relative”.  However this is only possible if there is a relative who is an employer or (possibly) self-employed. For every six Maori who were employed or self-employed, there is one relative assisting. In comparison, there were 132 non-Maori. Thus the Maori are more likely to employ a relative without pay, but have less opportunities.


Table 4 Maori in Non-Maori Quartile Income Ranges (By Occupation)


Non-Maori Quartile                0-25     25-50   50-75   75-100

Managers & Administrators    36.7     31.5     20.4     11.4

Professionals                           46.9     26.8     17.3     9.0

Technicians                             39.7     29.9     21.3     8.1

Clerks                                      35.9     28.4     24.1     11.6

Service & Sales Workers*       28.3     31.8     30.7     9.2

Agriculture & Fisheries           33.2     31.5     25.7     9.6

Trades Workers                       34.8     28.0     14.4     12.8

Operators & Assemblers         34.3     32.0     27.3     6.4

Elementary Occupations         30.3     32.7     29.8     7.2

TOTAL**                               38.4     33.5     21.6     6.5


Notes:  * including Armed Services    ** includes NAD.

Sources: “New Zealanders at Work” Table 11; “New Zealand Maori” Table 17.


In this section we are using income as a proxy for the quality of work within each occupation.  What we have done is calculate the quartile and median incomes for the non-Maori in each occupational group. Thus one quarter of the non-Maori are in each range. We have then calculated the proportion of Maori within each income range (Table 4). The perhaps not surprising result is that more than a quarter (typically a third) of Maori are in the lowest quartile while much less than a quarter (typically less than a tenth) are in the top quartile.


The point of this comparison is not to demonstrate that the Maori are on lower incomes, although that is true. Income is a crude approximation of job quality, for typically quality positions are well paid as well as having good working conditions, job security, and career prospects.  What the data supports is the proposition that not only is the Maori in less attractive occupational groups, but they have lower positions within the groups.




The conclusion from this data is that the Maori is in an inferior position in employment compared to the non-Maori, despite there being proportionally fewer Maori employed. We have suggested that the problem seems to be an occupational and skill one – for although there are differences between the industry deployment between the two groups, it is less obvious that this affects the relative job prospects.


On the other hand we find that the Maori are more common in those occupational measures which are unskilled, and for which there is a larger demand among the unemployed. Similarly on skill measures the Maori has less, less we suggested than the non-Maori who are a generation older. And the Maori are less likely to be in the top of the labour force, as employers, self-employed, managers, and professionals – the people who directly make employment decisions about who is to be employed, and whose wider compass of the labour market enables them to assist their relative and friends – their own social networks – to find positions.


We were unable to demonstrate from the data that the position of the Maori is improving relative to non-Maori, from generation to generation, although equally there is as little evidence it is worsening. There ware two reasons for this. First, and mainly, as explained in the chapter introduction, snapshot data cannot easily distinguish between the trajectory each cohort experiences through time, and the different trajectory each cohort experiences. On some issues we looked at a comparison between the 1991 and the 1986 census reports, but we found that sometimes data definitions had changed, and usually the data was too coarse to allow the identification of these relative changes.


If there is a policy conclusion in the previous paragraph it is that the gap between the Maori and the non-Maori may be narrowing in the younger cohorts – we cannot tell. What we can say is that any narrowing is small relative to the underlying size of the gap. If past policies have been working they have been working too slowly compared to the aspirations of the Maori and the nation.