QUANTITATIVE LITERATURE REVIEW AND PRELIMINARY FINDINGS
5.1 Introduction
In this section, factors effecting labour market outcomes with reference to the AIs, UKDs and ADs are explored. A selective review of the research in the area and preliminary empirical findings are presented. There are two main components to the section dealing with empirical findings. Firstly, the situation of the AIs in Australia is described using the complete ABS 1986 census data set. Secondly, the 1991 one percent population sample is used to conduct a more detailed statistical analysis, using a sample of AIs, Burghers and Anglo-Celanese to represent people in similar situation to AIs.
Specifically, issues such as labour force participation rates, unemployment rates and hourly income are explored. Particular attention is paid to how educational qualifications, occupation and industry employed in, effects unemployment and hourly income. The caste theory of (Ogbu, 1978; 1991) is explored using the techniques canvassed by Jones (1992a; 1992b).
Given the greater opportunity for upward social mobility in Australia compared to India, it becomes a point of interest to pose such questions as: "How are the AIs performing academically in Australia? What are their levels of academic and job attainment? Are AIs part of the so called "ethnic success ethic" (Bullivant, 1988: 238; Birrell and Khoo, 1995)? Or are they one of those migrant groups that does need assistance to perform up to expectation (Kalantzis and Cope, 1986; 1987). These questions will be explored further in this and following chapters.
5.2 Analysing the 1986 Census
The ABS makes available a tape file for researchers which contains data on one percent of the population. Unfortunately, when an attempt was made to extract a sample of respondents who were of European ancestry and were also born in India, an extremely small group of respondents was extracted. For this reason the assistance of the ABS was enlisted to provide information about all the AIs rather than a small subset.
The information provided by the ABS dealt with 18,000 AIs. Because of the large number of AI respondents in the census, the findings are more reliable than the 1991 findings.
5.2.1 Ancestry and the 1986 Census
The 1986 census asked respondents to indicate what their ancestry was. While the 1991 census and those prior to 1986 inquired only about whether a person was an Aboriginal or Torres Strait Islander, the 1986 census attempted to categorise the Australian population by self-described ancestry (ABS., 1984).
The ancestry question is vital when it comes to dealing with those groups of people who while retaining certain "ethnic" characteristics have lived away from their country of origin for hundreds of years. The Chinese and Indians are two groups which come readily to mind. When Chinese and Indians emigrate they tend to retain their language, religion and customs. As a result "ethnic" Chinese could emigrate to Australia from Malaysia while "ethnic" Indians could emigrate from Fiji.
The issue of ancestry becomes even more critical when attempting to analyse groups such as the AIs, who, while having a history going back over 400 years in India, do not view themselves and are not viewed as being "ethnic" Indians. Simply using country of birth as an ethnic marker for the AIs is inadequate to separate them from other Indians. Language and religion would help substantially in helping to distinguish AIs from other Indians, but it is the ancestry variable that is most likely to select out the AIs from other Indians.
In general, the ancestry question appears to provide reliable data. Jones (1991: 3) reports that data gathered through the ancestry question in the 1986 census, is valid and reliable, with only about 3.9 percent of immigrants describing their ancestry in terms that could not be easily categorised.
5.3 Analysing the 1991 Census
The 1991 census and the one percent sample produced by the ABS does not contain a general question on ancestry such as that in the 1986 Census. While this makes it more difficult to dis-aggregate the AIs from other Indians, the task is not impossible. The AIs are an English-speaking, Christian group from India, as such they can be separated out from other Indians by using the variables religion and language spoken at home. Once a data set containing AIs has been selected, then their labour market outcomes can be compared to other groups.
The analysis of immigrant labour market outcomes using statistical techniques has been a rapidly growing area of research in the last decade. The empirical methods of investigation adopted in this report are very much conditioned by previous work, in particular that of Jones (1992a; 1992b). The findings of previous research are summarised below so as to set the scene for the methods used and to provide the focus of our empirical analysis. The three key areas the present analysis will focus on will be participation rates, hourly wages and unemployment levels.
5.4 Participation in the Labour Force
In the present analysis, labour force participation rates represent the "labour force" (ABS, 1991: 76) expressed as a percentage of the civilian population aged 15 years and over. Labour force participation rates have been analysed by a number of researchers in an attempt to discern whether immigrants have lower participation rates than the Australian born and, if so, if this reflects a disadvantage. It is quite possible for a particular immigrant group to have low participation rates which can be explained by a high mean age for the group rather than disadvantage (Miller, 1986a).
Stricker and Sheahan (1979) attempted to identify the hidden unemployed by examining labour force participation rates. According to their findings immigrants were more likely to belong to the hidden unemployed, in that they had lower participation rates. Participation rates are strongly affected by the state of the economy. If the economy is doing poorly then workers are likely to become discouraged about their chances of getting a job and are less likely to try and enter the labour force. This is the so-called "discouraged ... worker effect" that lowers participation rates (Ackland and Williams, 1992: 2). There was also a stronger discouraged worker effect for immigrants from non-English-speaking countries (BLMR, 1986). Other researchers have found that immigrants do have lower participation rates when they first arrive in Australia but their participation rates quickly achieve levels that are similar to those of the Australian born (BLMR, 1986; Junankar and Pope, 1990). Further, any differences in participation rates between different birthplace groups have been explained by level of English proficiency and duration of residence in Australia (Brooks and Volker, 1983; Wooden and Robertson, 1989).
A caveat should be made at this point. Most research dealing with participation rates ignore the age structure of the immigrant and Australian born populations. Young (1992; 1995) has suggested that once participation rates are standardised for age then the participation rates of immigrants drop sharply relative to the Australian born.
In a recent study examining participation rates among Vietnamese, Maltese and Lebanese immigrants (Stromback et al., 1992: 32) it was found that participation rates for males actually decreased with years of education. This finding contrasted with earlier research that indicated increased participation rates with greater education. It appears that more highly educated male non-english speaking background (NESB) immigrants are more likely to stay out of the labour market. However, highly educated female immigrants tended to have high participation rates.
For both men and women participation rates increased with years of labour force experience in Australia. The above finding most probably reflects two issues. Firstly, a greater willingness on the part of employers to hire well educated women and secondly, women are probably more likely to accept work that men will not (Jones, 1992a).
Ackland and Williams (1992: 11, 19) have also noted the change in participation rates between the early 1980s and the early 1990s. In August of 1981 the overall labour force participation rate was 61.4 percent, with the overseas born having a higher participation rate of 62.9 percent, compared to 60.9 percent for those born in Australia. By August 1989 the position of the immigrants and Australian born had been reversed, with the Australian born having the higher participation rate of 64.6 percent and the immigrants having the lower participation rate of 61.5 percent.
5.4.1 Participation Rates and Year of Arrival in Australia
Young (1995: 41) not only confirms the above findings but suggests that immigrant participation rates are significantly lower than what most researchers report. This is because they fail to calculate "age-specific" participation rates.
Table 5.1 Year of Arrival in Australia for AIs and UKDs. 1991 Census, One Percent Sample. AIs UKDs % N % PRIOR TO 1971 156 27.2 6394 45.4 1971-1975 119 20.8 1414 10.0 1976-1980 43 7.5 622 4.4 1981-1985 57 9.9 758 5.4 1986-1987 29 5.1 312 2.2 1988-1989 40 7.0 459 3.3 1990-1991 16 2.8 489 3.5 NOT STATED 11 1.9 156 1.1 AUSTRALIAN BORN 102 17.8 3477 24.7 Total 573 100.0 14081 100.0
The majority of AIs arrived after 1970. This contrasts with the UKDs, most of whom arrived prior to 1971, see Table 5.1. Further, since there was a slowdown in immigration during the 1970s, the average age of immigrants would increase. This would result in a reduction in participation rates, since participation is higher for more recently arrived and younger groups.5.5 Unemployment
For the purposes of the present study a respondent was considered to be unemployed if they were looking for full or part time employment (ABS, 1991: 122). The issue of immigrant unemployment has received a great deal of attention from researchers. A number of researchers have found that the major factor explaining immigrants higher unemployment can be attributed to the settlement process, since it takes time to find a job and to restore the loss of skills or human capital experienced by migrating to Australia (Inglis and Stromback, 1983; Castles et. al., 1988). In this section the change in unemployment rates between 1981 and 1991 is explored along with factors such as age, education, and work experience, that affect unemployment.
5.5.1 Unemployment Rates 1981
The unemployment rates for the AIs in 1981 were relatively low. Considering that the AIs had only started arriving in Australia during the late sixties and early seventies, higher unemployment rates would have been expected.
Table 5.2 Unemployment rates for AIs, UKDs and ADs. 1981 Census.One Percent Sample. AI UKD AD Rate N Rate N Rate N Employed 93.3 220 92.4 5133 93.0 26488 Unemployed 5.6 13 7.6 419 7.0 1939 Total 100.0 233 100.0 5552 100.0 28491
New arrivals usually have much higher unemployment rates than groups that have been settled in Australia for long periods because it takes time to adjust to the new country and its ways (Inglis and Stromback, 1986; Miller, 1986b). Yet it is the established groups of UKDs and ADs that have the higher unemployment rates. This situation was to change by 1991.
5.5.2 Unemployment Rates 1991
By 1991 the unemployment figures for the AIs were now slightly higher than those for the UKDs and ADs, a complete reversal of the unemployment figures in 1981.
Table 5.3 Unemployment Rates for AIs, UKDs and ADs. 1991 Census One Percent Sample. AIs UKDs ADs Rate N Rate N Rate N Employed 88.27 301 88.67 6825 89.54 40632 Unemployed 11.73 40 11.33 872 10.46 4747 Total 100.00 341 100.00 7697 100.00 45379
The overall unemployment rate for AIs in 1991, at 11.7 percent, was just above that of the unemployment rate for the AKDs, 11.3 percent, and about a percentage point above the ADs, 10.5 percent. These figures were calculated using the ABS 1991 one percent population sub-sample and are higher than those that were obtained from the ABS labour force surveys. According to the ABS unemployment in August 1991 was 9.5 percent.
Table 5.4 Unemployment rates for AIs, UKDs and ADs, compared by age group. 1991 census one percent sample. AI UKD AD Age Rate N Rate N Rate N 15 - 19 15.4 13 27.3 297 21.6 4167 20 - 24 26.7 30 19.8 731 16.9 6084 25 - 34 12.0 83 10.6 1971 10.4 11893 35 - 44 8.5 94 9.0 2039 6.9 11771 45 - 54 7.6 79 7.5 1675 5.6 7485 55 - 59 16.0 25 10.5 550 8.5 1997 60+ 11.8 17 16.1 434 9.4 1982 TOTAL 11.7 341 11.3 7697 10.5 45376
The only age group at which AIs had lower unemployment levels than the UKDs and ADs was the 15 to 19 year olds. At almost every other age group the AIs had higher unemployment levels than the UKDs and ADs.Most immigrants arrive in Australia with experience in the labour force of their home country and overseas educational qualifications. Australian employers and professional associations who represent the "market" often fail to recognise these skills and qualifications as being useful in Australia (Iredale, 1987; 1988; Stromback et al. 1992: 60; Jones, 1992a; 1992b). On the positive side of the ledger, the proportion of time spent unemployed falls with length of residence in Australia (Harrison, 1984; Jones, 1992a; 1992b; McAllister, 1986).
One of the variables most strongly associated with unemployment is the newly arrived immigrant's ability to speak English. Studies that have included a measure of English proficiency have found that poor knowledge of English increases the chance of being unemployed (Gariano, 1993; Inglis and Stromback, 1986; Stromback et al., 1992; Jones, 1992a; 1992b).
In his analysis of 1991 census data Gariano (1993: 19) found that 32 percent of those born overseas with poor English language proficiency were unemployed. While this variable is included in many studies and is felt to be vital to explaining unemployment rates, it was not included in the present study, since AIs by definition are English speaking. Further, UKDs and ADs were included only if they indicated that they spoke English and no other language at home. This was to ensure that the three sub-populations being compared in the study were matched as closely as possible on English language proficiency.
Another factor which has been found to be important in accounting for immigrants' higher unemployment rates is their level of education. Research in general confirms that the higher a person's educational level the less likely he or she is to be unemployed. The issue becomes more complicated when the effect of overseas education on unemployment is introduced.
In a number of more recent studies (Beggs and Chapman, 1988a; 1988b; Jones, 1992a; 1992b), years of overseas schooling has been shown to have the effect of increasing unemployment levels. In contrast an actual qualification, whether a postgraduate qualification or certificate decreases unemployment levels for immigrants. Some studies have found that post-school qualifications gained in non-English speaking countries did not have as large an effect in reducing the chance of unemployment as the same types of qualification obtained in Australia (Wooden and Robertson, 1989).
Other studies have found that an overseas qualification substantially increases the likelihood of unemployment (Jones, 1992a; 1992b). A further caveat regarding overseas qualifications is that they may not be recognised in Australia which can make it more difficult for an immigrant worker to get a job (Iredale, 1987; 1988).
The factors discussed above are not the only factors that account for immigrants higher unemployment rates. There is little doubt that immigrants from particular countries and their descendants experience unemployment rates higher than can be explained by the above factors.
When the transferability gap, that is, the transfer of labour and educational skills, is large, as in the case of immigrants from non-English speaking countries, the problem of differential unemployment rates never really goes away. Thus some researchers are increasingly "more pessimistic" about the ability of certain immigrant groups to succeed in the labour market (Stromback et. al, 1992: 61). Some have argued that the high unemployment is due to the low skill levels associated with certain ethnic groups, in particular immigrants from non-English speaking countries (Harrison, 1984; Inglis and Stromback, 1983; Miller, 1985; Wooden and Robertson, 1989). To complicate the issue even further, Wooden and Robertson (1989) suggest that higher unemployment among Asians is caused mainly by the Vietnamese and those from the Middle East who are refugees and undoubtedly face a plethora of problems peculiar to refugees.
Ackland and Williams (1992: 29) in their study of labour force participation rates and unemployment levels conclude that for immigrants increases in the unemployment rate "... have been rising in absolute terms, and from larger initial levels". According to them, during the period 1974-1992 "...the effects of recessions on immigrants' labour market outcomes have been worsening".
The reason given for these higher unemployment levels is that the immigrants are increasingly likely to come from the Middle East and Asia rather than Europe. These immigrants are more likely to have problems with the English language and "Australian customs". While these explanations appear plausible for groups such as the Vietnamese, they provide little assistance in helping to explain what appears to be a substantial rise in AI unemployment rates between 1981 and 1991.
Having examined many of the issues that impact on a person's unemployment, the next section examines a particular model and its findings. It is this model that will be the basis of the econometric work to be conducted later in the study.
5.5.3 The Jones' Model of Unemployment and Findings
Jones (1992a; 1992b) studies are important with regard to the present work for a number of reasons. Firstly, they make a methodological and theoretical contribution to the field. Secondly, they use the complete 1986 census data set rather than the one percent samples that are available to most researchers (Chiswick and Miller, 1985; Stromback 1984; Beggs and Chapman, 1988a; 1988b).
The statistical model used to explain unemployment views unemployment as being primarily a function of schooling in Australia and overseas, educational qualifications and work experience in Australia and overseas. Unemployment is also seen as being indirectly related to marital status which is a proxy for differences in incentives to find a job. In the case of the present study religion was added to the model, and it was considered to be a proxy for skin colour, an issue to be explored in more detail later.
In a logistic regression, the predicted outcome of a particular event such as being unemployed "is the ratio of two probabilities converted to the scale of natural logarithms". In the case of predicting unemployment the dependent variable is the logarithm of the probability of being unemployed compared to that of being employed. When the odds are the same, there is a fifty-fifty chance of an event occurring, this logarithm is zero.
When the probability of unemployment is less likely than employment, which is true of most groups, their ratio is less than one, so the log-odds are negative. If the logistic regression parameters have negative signs, this indicates that the parameter reduces the risk of unemployment; if the parameters are positively signed, it indicates that the particular variable increases unemployment (Jones, 1992a : 46).
In an analysis of the 1986 census, where he compared Anglo-Celts, Dutch, Italians and Chinese, Jones (1992a: 49-52) reported the following results. Unemployment levels were lower among the Dutch and the Italians, but higher among the Chinese, who are a more recent and more highly qualified group. With the exception of Italians, women had a greater risk of being unemployed then men. Schooling and qualifications reduced unemployment, as did overseas labour force experience (but not as much as local experience).
For males, the Dutch and Italians had substantially lower unemployment rates than the Anglo-Celts, but then they also had many more years of Australian labour force experience. The male Chinese in contrast had appreciably higher unemployment rates and many fewer years of Australian labour force experience (Jones, 1992a: 37-40).
When Jones (1992a) controlled for the effects of schooling, qualifications and Australian labour force experience the unemployment situation did not change at all. The Dutch had the lowest unemployment followed by the Anglo-Celts, Italians, Chinese and then a sub-group of Chinese, those from Vietnam. For women, the model projections virtually reversed themselves. It was now Italian women who had the highest unemployment levels, followed by the Dutch, Anglo-Celts, Chinese from Vietnam and lastly other Chinese. Jones (1992a: 60) suggests that a possible reason for the Chinese and Vietnamese women having low unemployment rates could be due to their acceptance of "low status and poorly rewarded jobs".
A man who is married is usually expected to "go out and earn a living" rather than relying on the job search allowance to maintain himself and his family. As a consequence he may be willing to except any available job and so reduce unemployment for married men. Further, employers are more likely to keep on a married man than an unmarried one because he has "a family to support."
The position could be different for a woman, in that many conservative employers could take the view that a married woman can "fall back" on their husbands' wages and so they might be more willing to dismiss a married woman during times of economic recession (Pech, 1991). Other employers may prefer married women because they already have experience in the workforce and often are willing to work short hours so that they can spend time looking after their families. In other cases employers prefer younger women workers who, because they lack experience, they can pay less (Jones, 1992a: 45).
In general, married people have a lower risk of being unemployed. This also reflects the fact that they are older, on average, than the unmarried and as a result have saleable skills. Overseas qualifications provide weaker protection against unemployment than qualifications obtained after arrival in Australia. A single year of schooling has an effect more than ten times larger than a single year of experience, with stronger effects for Australian labour force experience rather than overseas experience (Jones, 1992a; 1992b).
To conclude, while years of schooling and qualifications reduce unemployment, overseas schooling and qualifications have a substantially weaker effect. Further, while overseas labour force experience helps reduce unemployment it does not have as strong an effect as that of Australian labour force experience (Jones, 1992a: 110).
5.6 Hourly Earnings
After unemployment, the most important issue for immigrants is the level of their earnings. Most econometric studies examine the effect of a number of predictor variables on a person's hourly income as one of the main aspects of their research strategy (Haig, 1980; Jones, 1992a; 1992b; Stromback, 1984).
The first econometric analysis of earnings conducted in Australia was that by Haig (1980). He tried to discover if there was any evidence that migrants were being discriminated against by using data collected for the Henderson Inquiry into Poverty in August 1973. He used a number of variables such as ethnic background and length of time spent in Australia to estimate regression equations.
According to Haig (1980) workers from overseas earned about 6 percent more than those born in Australia. The main reasons for this wage advantage were that they worked longer hours, were better qualified, tended to be older, male and concentrated in the urban areas. In Haig's opinion immigrant workers should have had hourly earnings that were 9 percent higher than that of the Australian born given their superior endowments of labour force experience and education. Haig ascribed some of the 3 percent loss in earnings that immigrants experienced partly to the lack of recognition of overseas qualifications.
Chapman and Miller (1983) analysed the 1976 census using an econometric approach where they found that Australian-born persons received higher rates of return than immigrants for their education and labour skills. For those who were Australian born, each year of education increased earnings by an extra 4 per cent for males and 3.5 per cent for females, while those with greater labour market experience increased their earnings by more than 2 percent for males and 5 percent for females for each additional year of labour force experience.
Mulvey (1986; cited in Stromback et al. 1992: 95) points out substantial differences between different birthplace groups. According to Mulvey (1986), compared to people born in "other" countries men and women born in the UK received 6.18 percent higher hourly wages, net of a host of individual characteristics. These characteristics included geographic location, years of schooling, amount of labour market experience and marital status.
Given Mulvey's findings that people from the U.K do relatively well compared to immigrants from other countries, UKDs appear to be a group that will provide a useful comparison group for the AIs. Both groups are Christian, English speaking and Westernised. The main difference is that AIs have an Asian heritage. As such they will differ from UKDs on some cultural issues and in many cases colour.
Many of the econometric studies have limitations of one type or another. For example the Chapman and Miller (1983) study aggregated data across very broad categories such as overseas-born versus Australian-born. No separate adjustments were made for labour force experience in the country of origin and then again in Australia.
The overseas-born/Australian dichotomy hides between-country differences which are extremely important given variations in language proficiency among the overseas born, which was not controlled. Further, formal education was represented only by the number of years of schooling, with no dummy variables for having a degree or postgraduate qualification were included. Similarly Mulvey (1986) did not distinguish between pre- and post-migration experience, or identify the consequences on wages of English language proficiency.
During the 1980s and 1990s econometric studies have used the one in 100 samples provided by the ABS, which allows the researcher to develop regression models based on relatively large numbers of respondents (Stromback, 1984; Chiswick and Miller, 1985; Beggs and Chapman, 1988a; 1988b).
A study (Stromback, 1984) that analysed the 1981 one percent ABS data set found immigrants from English-speaking countries tend to receive similar rates of return to those of the Australian-born with regard to education and labour force experience. On the other hand those from non-English speaking countries tend to achieve substantially lower rates of return for their educational and labour force experience. This could reflect non-recognition of overseas qualifications and overseas experience. These findings were similar to another study conducted by Tran-Nam and Nevile (1988).
Similarly, Cheswick and Miller (1985) found that, in general, male immigrants received lower returns to home-country education and experience than they did for education and experience acquired in Australia. Beggs and Chapman (1988b), using a non-parametric approach to estimate hourly wages, found that male immigrants with low levels of measured education receive higher hourly income than do the equivalently educated Australian-born. But as immigrants' measured level of education increases, their wage position deteriorates relative to the equivalently qualified Australian-born. It appears that immigrants with high educational qualifications do relatively poorly, in terms of income. This result found also in Wooden and Robertson (1989).
A possible explanation for this phenomenon is that Australian schooling is of higher quality than overseas schooling (Evans and Kelley, 1986). The Beggs and Chapman (1988) model did not allow a comparison between Australian and overseas schooling with regard to income. But given that some part of an immigrants' education was not received in Australia their results are consistent with this theory.
A second possible explanation for why immigrants with high educational qualifications often do relatively poorly could be due to discrimination. As the Annual Report (1989-1990) of the Public Service Commissioner shows, first and second generation immigrants were less likely to reach the middle and top levels of the public service "given their qualifications and length of service" (cited in Flatau and Hemmings, 1991: 52). Similar findings have been made by Watson (1995) in a more recent report.
Chapman and Iredale (1990) provided more support for the propositions developed in Beggs and Chapman (1988a; 1988b), namely that as immigrants' level of education increased, so too did their wage disadvantage relative to similarly educated Australian-born persons. As far as wages are concerned, immigrants from NESB with high levels of skills obtained overseas fared relatively poorly in the Australian labour market. The qualifications with the lowest returns were higher degrees for males, certificates/diplomas for males, and trade certificates for both males and females.
More recently (Watson, 1995) has provided evidence that immigrants from a non-English-speaking background often fail to gain an adequate share of managerial positions in Australia. There now appears to be evidence from a number of different studies that there is some discrimination in the Australia work place. This issue will be explored further in the next section.
5.7 The Jones Model of Hourly Wages
Jones (1992a; 1992b) in a recent study examining the role of ethnic background in the Australian labour market used what is now a common methodological approach, namely building regression models for different groups. He used Ordinary Least Squares (OLS) models to predict hourly earnings for males and females within four ancestry groups, while holding constant a number of factors. The four ancestry groups were the Anglo-Celts, Dutch, Italians and Chinese.
The main difference between Jones' study and other studies of its type in Australia was that Jones has access to the total census data set, rather than having to deal with a relatively small sample, such as that provided by the ABS 1 in 100 samples. Further, rather than relying on birthplace as a marker of ethnicity he was able to introduce the concept of ancestry into his study.
The 1986 Census included a question on a person's ancestry. So Jones (1992a; 1992b) was able to group together not only people who had been born in a particular country, such as Italy, but also those who were descended from Italians, but had been born in Australia.
Jones compared Anglo-Celts, Dutch, Italians and Chinese. The Anglo-Celts, of course, represent the core or majority grouping, while both the Dutch and Italians are relatively well-established European groups. The Chinese represent a new, Asian group, even though there have been small numbers of Chinese since the establishment of Australia as an British outpost.
When Jones (1992a) examined the issue of hourly earnings he found that Anglo-Celts earned most, followed by the Dutch, Chinese and then the Italians. Once Jones adjusted for Australian labour force experience, years of schooling and qualification the results changed. Both male and female Anglo-Celts were penalised for having few years of schooling, in that they earned less than the other groups, the exception being the female Italians who earned even less. But once the Anglo-Celts achieved trade or degree qualifications they tended to earn substantially more than the other groups (Jones, 1992a: 93).
Like other recent studies, Jones (1992a: 114) found that non-Anglo-Celtic groups "get low returns for their overseas schooling". Further, the difference in years of schooling does not explain as much of the variation in hourly income among the minority groups as it does among the Anglo-Celts (Jones, 1992a: 110). Women earn less than men and have higher unemployment levels mainly because, in the opinion of some researchers, their experience and qualifications are not recognised and they have to contend with a glass ceiling (England, 1979; Jones, 1992a: 113)
Econometric research that has attempted to explore the possibility of discrimination against a minority group has usually compared the performance of the minority group on a number of variables to the majority grouping. If a member of a minority group who is similarly qualified and experienced to a member of the majority group earns less and experiences a higher risk of unemployment, we could then infer the presence of discrimination. The problem of course is to allow for those group differences which have a significant impact on a person's job performance. The members of a particular minority group may be grouped in jobs with lower incomes, but this may be due to a lack of relevant work experience rather than discrimination (Jones, 1992a: 14).
Factors other than discrimination and human capital contribute to differences between groups. There are factors such as measurement error involved in variables such as hourly income. Selectivity bias is another problem, where some workers may not seek employment in particular industries if they anticipate discrimination. Further problems may be caused by omitted variables such as length of tenure in present job, on-the-job training, supervisory responsibilities, and size of firm.
Most of the econometric studies of inequality include only a small subset of the many variables that influence how much workers get paid and their likelihood of being unemployed. So like most areas of social science, the econometric approach has its limitations (Jones, 1992a: 15).
In spite of its imperfections, the econometric, human capital approach is probably the approach that has gained the most acceptance among researchers. When it comes to answering the question of whether or not groups with comparable characteristics receive similar pay rates and have similar unemployment rates, it is the econometric approach that appears to be the most useful (Jones, 1992a: 16).
While the present study attempts to examine the economic performance of AIs in Australia, it is virtually impossible to take into account the effects of at least two hundred years of discrimination and prejudice by the British in India and perhaps 400 years of discrimination by the Indians. An Indian Christian cleric based in Perth, notes that young AIs hardly ever "... go in for tertiary education", further he suggests that those AIs who start their Australian education at high school level "... find it rather difficult to adjust... (Rivett, 1975: 205)."
Ogbu (1978; 1991) argues that Afro-Americans are failing to achieve educationally and job-wise because of the destructive effects of racism over hundreds of years. Similarly Jones (1992a: 16) suggests the possibility that Australian Aborigines may opt out of schooling early because they do not believe that they will have a fair chance in the job market.
As a result of discrimination or their own inability to communicate in English, members of ethnic groups who cannot get a job often start their own businesses, while others try and put up with the stresses and pressures of being unemployed (Evans, 1989; Tait et al., 1989: 192; Pascoe, 1990: 1-4; Jones, 1992a: 17).
Previous studies have found that the Australian labour market penalises weak English-language competence. In fact this is probably the variable more than any other that affects the job performance of immigrants. Every-day experience suggests that immigrants who speak English well will perform well in the Australian market place. Further, research strongly supports the above conclusion (Jones, 1992a: 50-52, 77-79; 1992b: 140-141, 147-148). AIs, given that English is their mother tongue, have a huge advantage over many Asian and European immigrants, who experience difficulty because "low English proficiency ... hamper[s] their socio-economic progress (Stromback et al., 1992: 12)." The Australian labour market devalues overseas schooling and labour force experience. As a consequence, immigrants suffer higher rates of unemployment and receive less pay than they might otherwise expect.
5.8 The Methodology Used to Analyse the 1991 Census
The methodology adopted in most econometric studies, including this one, involves estimating regression equations and comparing how different groups fare in the labour market. Particular emphasis is placed on hourly earnings and unemployment, relative to the human capital, that is their qualifications and labour force experience, possessed by immigrants. The more human capital, in the form of qualifications and work skills, workers supply to the labour market, the better they can expect to do relative to other workers.
The strength of the relationship between work skills, qualifications and earnings may vary across the labour market, and between groups. The present study while focusing primarily on the Australia wide performance of AIs, UKDs and ADs does not ignore the possibility of regional variations. For example, the better educated and more highly skilled AI workers are more likely to find work in the larger cities of Melbourne and Sydney rather than in Adelaide or Perth.
The AIs are a group that has had to cope with discrimination for a long period in India, with many commentators suggesting that their academic and job attainment is poor. The econometric analysis that follows attempts to assess the AI socio-economic attainment in Australia. The emphasis on discrimination in this chapter is directly related to the previous chapters where the long term consequences of discrimination are discussed. The second and fourth chapters in this study established the status of AIs in India. The present chapter attempts to establish the status of AIs in Australia.
5.9 Model Description
What appears below is a detailed discussion of the variables used to develop the regression models used in the present study. Table 5.5 provides the reader with the variable names used in the statistical analysis and a description of what each variable label stands for.
The variables below provide the basis for the analyses that follow. Throughout this part of the study people of Australian descent (AD) - that is people with both parents born in Australia and who speak only English at home - represent the comparison or reference group. The vast majority of these Australians will be of Anglo-Celtic origin, that is persons of Australian, British or Irish descent.
The main question that the present analysis attempts to answer is how well AIs are doing relative to members of the main AD grouping. Further, the present analysis attempts to discover whether any observed differences are consistent with possible discrimination experienced by the AIs in India and in Australia. Having experienced discrimination for many generations in India the AIs may have altered their behaviour to cope with lowered expectations, a process that may have occurred with the Aborigines in Australia (Jones, 1992a: 16) or with the Afro-Americans (Ogbu, 1978; 1991).
Table 5.5 Model Variable definitions Variable Description MARRIED Dummy variable. One if the respondent is married HIDEGREE Dummy variable. One if the respondent has a Postgraduate qualification; DEGREE One if the respondents highest qualification was a degree; DIPLOMA Dummy variable. One if the respondents highest qualification was a diploma; TRADEQ Dummy variable. One if the respondents highest qualification was in the trades; MANAGER Dummy variable. One if the respondent was in the occupational category of manager; PROF Dummy variable. One if the respondent was in the occupational categories of professional para-professional; CLERSAL Dummy variable. One if the respondent was in the occupational categories of clerical. sales or services; TRADES Dummy variable. One if the respondent was in the trades, skilled agriculture, plant operating; SECONDRY Dummy variable. One if the respondent was in the industry categories of manufacturing construction and electricity, gas and water; SERVICE Dummy variable. One if the respondent was in the wholesale and retail trade, transport communication and recreation; FINANCE Dummy variable. One if the respondent was in the industry categories of finance. property and business; FEDGOVT Dummy variable. One if the respondent was employed by the Federal Government; STATEGVT Dummy variable. One if the respondent was employed by the State Government; FEMALE Dummy variable. One if respondent is female; NSW Dummy variable. One if respondent is from NSW; VIC Dummy variable. One if the respondent is from Victoria; PERTH Dummy variable. One if the respondent is from Perth; CATHOLIC Dummy variable. One if the respondent was Catholic; ANGLICAN Dummy variable. One if the respondent was Anglican; FORGNED Continuous variable. Years of overseas education; AUSTED Continuous variable. Years of Australian education; AUSLFE Continuous variable. Years of Australian labour force experience; BORNAUST Dummy variable. One if born in Australia; HRS49 Dummy variable. One if the respondent worked 49 or more hours per week; AT16 Dummy Variable. One if the respondent arrived in Australia at age 16 years or younger; FORGNLFE Continuous variable. Years of overseas labour force experience; SELF_EMP Dummy variable. One if the respondent is self employed; BORNASIA Dummy variable. One if the respondent was born in Asia; FOREIGNQ Dummy variable. One if the respondent had an overseas qualification; Dependent Variable HRINC Hourly income in dollars per hour.
In comparing labour force participation rates, unemployment and hourly income between different groups, two variables were of primary interest, education and labour force experience that was gained both in Australia and overseas. It is essential to distinguish education and labour force experience acquired before and after immigration to Australia, simply because Australian employers seem to treat them differently (Jones, 1992a: 27).
5.10 The Variables Used
The 1991 Census one percent population sub-sample includes a large number of variables which includes information on a person's age, age left school and period of residence in Australia. These variables provide enough information to calculate years of education completed before and after arrival in Australia, as well as years of overseas and Australian labour force experience.
Having only a high school education would handicap a person's earnings and employment opportunities, while having a post-school qualification, would improve a person's income and employment chances (Jones, 1992a; 1992b). But there is a caveat here. Australian employers do not recognise overseas qualifications from many Asian and European countries. Further, an overseas qualification often does not confer as much of an advantage in the work place as an Australian qualification, particularly if it is from Asia.
In Australia most employers are heavily influenced by "formal credentials" when assessing a job applicant (Broom et al., 1980: 74-88; Jones, 1992a: 28; 1992b). In the regression analyses that will discussed in chapter seven, not only were overseas and Australian years of education entered but also a dummy variable representing overseas qualifications.
5.11 Adjustments to Labour Force Experience for Women
When estimating labour force experience from census data the normal procedure is to subtract the person's age at when they completed full-time education from their age at the time of the census. While this approach works reasonably for men, it is problematic when used for women. Most women move in and out of the work force depending upon which stage they are in the child bearing and rearing process. Most women start off working full-time, then stop working for a while to have children and then gradually return to the part- and full-time workforce. Jones (1992b: 125) has suggested a method by which it is possible to adjust for this "intermittent labour force activity" of women.
Jones (1992b: 125-129) used the National Social Science Survey (NSSS; Kelley and Bean, 1988) results on occupational histories to adjust for differences in labour force experience between men and women. The adjustment for part-time and intermittent work hardly affected the estimates of labour force experience for men. For women, however, it reduced the usual measure by as much as one-third. It should be pointed out these adjustments were based on aggregate data. Men and women from different ethnic backgrounds may have lower, or higher, participation rates than women in general (Evans, 1984). While Jones (1992b: 130) concedes that these adjustments are unlikely to be "optimal", he points out that "some adjustment is better than none". For a worked example of this labour force adjustment for women see Jones (1992b: 152).
Data from the (NSSS) were used to adjust for the differences between men and women regarding part-, full-time and intermittent work. The survey asked respondents to nominate, first, for how many years they had worked most of the year full-time, and second for how may years they had worked most of the year part-time. Responses and an age/gender breakdown appear below, in Table 5.6.
Table 5.6 Breakdown by Age and Gender for percentage of years in workforce spent working Full-time. (Jones, 1992b: 127) Age-Group Gender % of Years Working Average Years in Full-Time Labour Force
Under 25 Male 88.5 4.40 Female 89.8 4.21 25-34 Male 94.6 11.00 Female 83.7 9.19 35-44 Male 98.0 20.85 Female 71.9 16.65 45-54 Male 98.6 31.91 Female 73.1 21.24 55-64 Male 94.8 39.95 Female 67.6 26.37
Except for those under 25 years of age, women are more likely to work part-time than men. Women are also more likely to have intermittent labour force careers, as shown by their lower years in the labour force averages. By the time men and women are in the 55-64 age group this difference has increased to 13.58 years. Two regressions were conducted, one relating the total number of years worked full-time to two variables, age and a quadratic, age-squared, to allow for any curve-linearity in the relationship. This regression gave the results plotted in Figure 5.1. The second regression related the percentage of years worked full-time to age and a quadratic, age-squared, once again to allow for any curve-linearity in the relationships. The results of this regression are plotted in Figure 5.2 by single years of age.
Figure 5.1, which graphs the estimated number of years spent working for pay most of the year since finishing education, shows that women work more than men up to the age of 23. After the age of 23, the older the two groups get, the greater the number of years men are likely to have worked relative to women. In Figure 5.2, there is an interesting curvilinear relationship between age and percentage of years worked full-time. At the age of 23 both men and women are likely to have spent about 90 percent of their working life working full-time. But, by about the age of 50, men have spent about 2 percent of their working lives working part-time compared to 30 percent for women.
Figure 5.1 Estimated Years Spent Working for Pay Most of the Year since Finishing Education (Jones, 1992b: 128).
Figure 5.2 Estimated Percentage of Working Life Spent Working Full-time (Jones, 1992b: 128).
Since the census contains no information on a persons occupational history, the estimates shown in Figures 5.1 and 5.2 will be used to adjust labour force experience for the effects of part-time and intermittent work. The results shown in Figure 5.1 allow the calculation of a female/male ratio reflecting the difference in average years spent working by men and women of the same age. The regression results plotted in Figure 5.2 provide a basis for adjusting potential labour force experience for part-time work, by estimating the relative proportion of a person's working life spent on part-time rather than full-time work.
Except for women under the age of 23, this adjustment reduces their labour force experience relative to men. The estimated effects of labour force experience on different outcomes for men are hardly affected at all by this adjustment, this finding was made by Jones (1992a; 1992b) in his studies and also in the present study. For women the situation was quite different, their labour force experience rates are substantially higher than their adjusted labour force experience, as is to be expected from figures 5.1 and 5.2.
5.12 Calculating Hourly Income
The Australian census provides information that may be used to calculate unemployment and hourly income. A variable that deals with labour force status provides the researcher with information about which persons are unemployed and whether they are looking for full-time or part-time employment, which people are employers, self-employed, employees and unpaid helpers.
Weekly income was calculated by dividing the individual's income over the period of a year by 50, the average number of weeks worked by most people. The weekly income was divided by the number of hours the person worked on a weekly basis to give hourly income. Because the census question on income sought information only in broad categories, category mid-points were used to calculate hourly income. The exception was for the highest open-ended category, where an estimate of $50,000 was used.
Hours worked was another variable that was provided in categories and once again mid-points were used to perform the hourly wage calculations. While Jones (1992a: 29) used 60 hours as an estimate for the final category of 49 or more hours worked a week, the present study used an estimate of 50 hours. Although the census reports income from all sources, the analyses in the present study treat annual income as occupational earnings.
Only full-time workers are included in the earnings analysis because people who work part-time are more likely to work for only part of the year, which would distort estimates of hourly earnings. Selecting only those workers who have indicated that they worked 35 or more hours per week does result in a reduction of female workers since many more women than men work part-time (Jones, 1992a: 87).
5.13 The Basis for the Group Comparisons
The present study has concerned itself primarily with the following measures of human capital. Firstly, years of education and educational qualifications and secondly, labour force experience both in Australia and overseas. A third issue, regional variations were also explored, since AIs might be treated better by the market in certain regions rather than others.
If an employer rewards his workers by giving them promotions and pay rises on the basis of other than work related performance, this practice may be viewed as being discriminatory. By comparing people from different ethnic backgrounds, but with similar educational and job profiles, the researcher can discover possible discrimination (Evans and Kelley 1986: 188-189; Flatau and Hemmings, 1991; Jones, 1992a; 1992b; Jones and Kelley, 1984; Marini, 1989: 347; Stromback, 1984).
The human capital approach, combined with statistical techniques commonly used in econometrics, allows the researcher to identify discrimination in the market place. Discrimination is identified in terms of different levels of income for different groups while attempting to make sure, statistically, that these groups are comparable so that like is being compared with like.
The following analyses compare members of two immigrant groups, the AIs and UKDs with members from the majority or "control grouping", ADs. Groups that fare worse than ADs represent examples of negative discrimination, just as groups that fare better are possible cases of positive discrimination.
It is possible that immigrant Anglo-Celts or UKDs do better than ADs because Australian employers believe that their British qualifications and job skills are better than those of ADs and other people from overseas such as the AIs. As Jones (1992b: 131) has suggested, this is perhaps "an echo of the cultural cringe" from the days when Australia was quite definitely a British colony.
There is a simple way to make group comparisons using OLS models (Jones and Kelley, 1984). It involves substituting the human capital characteristics of one group with the regression equation of another. For example, to model the hourly earnings of AIs given that they had the human capital of ADs, the AI means would be replaced with the means of ADs and the new regression line would be plotted, see (Jones, 1992b: 142). This new regression line provides an insight into how AIs would fare if they had the human capital of ADs.
In Chapter seven, the AIs have their human capital substituted with that of ADs, the UKDs have their human capital substituted by that of ADs and the ADs have their human capital substituted by that of the AIs. This last substitution was meant to provide an insight into how ADs would perform if the market treated them like AIs.
5.14 Discrimination
The present study has as its main hypothesis that the AIs in India were and are a caste group. Many AIs have immigrated from India to Australia and conceivably have transformed their caste mind-set, because of the process of immigration. The purpose of exploring the issue of discrimination in Australia is to evaluate the view that the AIs remain a caste group in Australia. On the other hand, the evidence may support the opposite hypothesis, that is the AIs in Australia are attaining well, similar to most other immigrant groups.
A number of researchers have shown that some immigrant groups have a high proportion of degree holders (Birrell, 1987; 1994; Birrell and Khoo, 1995) and that in many cases the children of immigrants are more likely than the children of ADs to complete high school (Martin and Meade, 1979; Meade, 1983a; 1983b) and to go on to University (Williams, 1987; Williams et al., 1993). The work of these researchers suggests that there is little evidence that immigrants or their children are discriminated against, either in their educational opportunities and attainment or in the professional arena (Birrell and Khoo, 1995).
The work of other researchers have suggested that these findings need to be qualified. It has been consistently found that overseas qualifications are often not recognised in Australia. The formal recognition of overseas qualifications by a professional or governmental body being only part of the problem. In general it is the "market" or employers that fails to place much value on overseas education (Stromback et. al. 1992: 61). Most employers show a strong preference for Australian labour force experience and Australian educational qualifications (Iredale, 1987; 1988).
Stromback et. al (1992) in their comparison of a number of different birthplace groups concluded that there is a lower rate of return for Australian education received by the Maltese, and to a lesser extent the Lebanese, which in their opinion suggested some evidence that Australian employers do engage in discrimination. Further, the tendency for higher unemployment rates and lower participation rates among the Vietnamese and Lebanese suggest the possibility that the Vietnamese and Lebanese are discriminated against.
The finding of higher unemployment rates and lower participation rates was obtained even when refugee status was controlled for. The difficulties these immigrants experience in finding jobs cannot be explained only by their refugee status. While there are a number of factors common to these immigrants that contribute to their difficulties, statistically controlling these factors allows the researcher to compare like with like, providing evidence for the possibility of discrimination (Stromback et. al. 1992: 61).
Much of the early research into immigrants' labour market experiences suggested that immigrants did reasonably well in the Australian labour market after allowing for recency of arrival and low English proficiency (Blandy et. al. 1979). However, some of the more recent research findings have painted a more pessimistic picture (Annual Report '1989-1990' of the Public Service Commissioner; cited in Flatau and Hemmings, 1991: 52; BLMR, 1986; Chiswick and Miller, 1985; Miller, 1985: 6; Tran-Nam and Nevile, 1988; Watson, 1995). In part this is due to the restructuring of the economy and the resulting poor labour market situation, which, while having an adverse effect on all Australians, has affected recent immigrants disproportionately.
In general the labour market outcomes for newly arrived migrants are poor if not "dismal". Those immigrants who arrived when jobs were easier to get, such as the Maltese, do continue to have relatively low unemployment levels but have not been able to advance socio-economically. Thus, there is the possibility that employers are discriminating against people based on their ethnic background. If the long-term performance of the Maltese is any guide, the prospects for relative newcomers such as the Vietnamese and Lebanese are unlikely to be any better (Stromback et. al. 1992: 61).
5.15 Chapter Summary
This chapter's primary concern was with reviewing the econometric Australian literature dealing with labour force participation rates, unemployment and hourly earnings. Further the 1986 censes with its ancestry question was also discussed along with some of the gender issues related to participating in the labour force.
To briefly summarise the literature review, most past research indicates that immigrants have higher participation rates compared to the Australian born. Recently, however, studies that have adjusted participation rates for age indicate that immigrants in general display lower labour force participation rates than the Australia born. At the same time, unemployment rates for immigrants have been rising in recent years. This is primarily because of the more difficult economic circumstances that have befallen Australia during the 1980s and 1990s. With regard to hourly income, most research indicates that immigrants earn similarly to the Australian born.
