Unit Record Data Research Articles

Do Lower Expected Wage Benefits Explain Ethnic Gaps in Job-related Training? Evidence from New Zealand

Many studies show that individuals from ethnic minority groups receive low levels of job-related training, raising the question of whether lower expected wage benefits contribute to this lack of training. In this paper, unit record data are used to examine the effect of job- related training on wages in New Zealand. The results suggest that both the receipt of employer-provided training, and the number of training events, have larger effects on wages for minority workers than they do for white workers. There are no differences across ethnic groups in the wage benefits from other types of training.

Births, Infants and Children: An Econometric Portrait of Women and Children in India

This article undertakes an econometric analysis of the constellation of factors that serve to determine some outcomes with respect to demography and to schooling in India. These are: the numbers of pregnancies, live births and infant survivals to women, the chances of children being enrolled at school and, if enrolled, of continuing in school. The econometric estimates are based on unit record data from a survey — carried out by the National Council of Applied Economic Research, New Delhi — of 33,000 rural households in sixteen states of India. The study concludes that a broad spectrum of factors affects these outcomes. The literacy of women is important but so is the literacy of men. Infrastructure, in the form of safe drinking water and easy access to medical facilities, is important for infant survivals and, in the shape of easy access to schools, is important for school enrolment. Parental occupation matters for both infant survivals and schooling: children born to women who work as labourers are disadvantaged, relative to other children, in terms of their chances of surviving infancy and, if they do survive, of receiving schooling. The number of siblings that a child has affects his/her schooling outcomes, and gender, religion and region play an important role.

DOMESTIC LEISURE TRAVELER PURCHASE AND CONSUMPTION SYSTEMS

Leisure travelers visiting overnight destinations complete a series of related purchases that may be viewed usefully as a leisure travel consumption system. Certain travel-related purchases in one product-service category trigger additional purchases in other product-service categories. This article reports findings of an empirical study that support the above core proposition. For this study, analyses were conducted on the unit record data of overnight domestic travelers (n = 27,653 households) in the Australian National Visitor Survey (NVS) 1998 study on the characteristics and travel patterns of domestic tourists. A core proposition for doing the research work reported here is that understanding domestic overnight travel behavior requires examining tourism consumption systems rather than only focusing on two or three variable relationships. The theoretical logic of focusing on tourism trips as a unit of analysis rather than variable relationships is applicable through field studies of tourism consumption systems.

The Effect of Education on the Timing of Marriage and First Birth in Pakistan

A rapid rise in women’s education levels, an increase in the age at marriage and an increase in the age at which they have their first child are key features of demographic transition in any country. Education is considered to be an essential component in this process because increases in educational attainment are likely to significantly affect both age at marriage and the duration between marriage and first birth - in particular increasing both the age at marriage and the time to first child. This paper uses individual level unit record data from Pakistan to examine the effect of education on the age at marriage and on the duration between marriage and first birth. We jointly estimate educational attainment, age at marriage and duration between marriage and first birth allowing for household level unobserved heterogeneity. We find that educational attainment increases the age at marriage but does not have a statistically significant effect on the duration between marriage and first birth. However, there is a threshold level of education that must be attained before educational attainment starts having an effect on the hazard of marriage.

International Poverty Comparisons on Unit Record Data of Developing and Developed Countries

Cross country poverty comparisons on unit records have, rarely, involved both developing and developed countries. The present study attempts to fill this gap by comparing poverty across fourteen nations with diverse economic and demographic characteristics and at vastly different stages of economic development. The study contains evidence on (a) cross country variation in the equivalence scales estimated in the presence of both household size economies and adult/child relativities, (b) impact of demographic adjustment of the poverty line, that incorporate household size and composition changes, on the poverty rates, and (c) sensitivity of the poverty estimates and their rankings to the ‘equivalence elasticity’. The study finds that country rankings based on per capita GNP bear very little resemblance with that based on the aggregate poverty rates. The latter hide substantial variation in the poverty estimates across different household types.

A cohort analysis of household income, consumption and saving

This paper presents a cohort analysis of household income, consumption and saving in New Zealand. It is based on an analysis of unit record data from March years 1984 to 1998 taken from the Household Economic Survey (HES). These data are a series of cross‐sectional surveys rather than a true panel, so we construct synthetic cohorts and use a range of regression models to separate out the effect of age, birth‐year cohort and survey year on income, consumption and saving rates. There appears to be a “V” shaped cohort pattern in household income and saving, such that the age profile of saving shifted down for the cohorts born between 1920 and 1939 relative to the younger and older cohorts studied.

Two Aspects of Fertility Behavior in South Africa

In this article we analyze two aspects of female fertility using unit record data from South Africa. We examine the effect of individual, household, and community characteristics on the age at first conception and the number of pregnancies for a national sample of women. The segmented nature of theeconomy of South Africa allows us to study differences in fertility patterns among the different socioeconomic classes and across the different races. 1 While much has been said about fertility behavior of women in sub-Saharan Africa, an analysis of the fertility patterns of South African women is significantly absent. 2 In this article we attempt to fill this gap in the literature. The first aspect of fertility behavior that we examine is the age at first conception. In addition to the significant biological components associated with the age at first conception there are economic and nonbiological factors that affect the probability of conceiving by influencing the choices that women make (including choices regarding birth control and other family-planning strategies). Age at first conception is modeled using a hazard model. Of particular importance is the effect of the highest level of education attained on the age at first conception. The estimation results show that the more educated a woman is, the higher is the age at first conception subject to a threshold level of education. There are no significant racial differences in the age at first conception. The second aspect of fertility examined is the number of times a woman has been pregnant. Economic theories of fertility assumethatparentsoptimally choose the number of children they want to have, taking into account the costs and benefits associated with a given number of births, including the costs of avoiding unwanted births. Assuming that no pregnancy is unwanted, the number of times a woman is pregnant over her reproductive age can

Cervical screening by socio-economic status in Australia.

To examine differentials and time trends in self-reported Pap test rates by socio-economic status (SES) from the 1989/90 and 1995 Australian National Health Surveys (NHS). The unit record data for females were extracted from the two NHSs and combined. The outcome variable of interest was 'having a Pap test in the past three years'. The principal study factor was SES measured as individual characteristics and SES of area of residence. Migrant status, rurality, year of survey and age were controlled for in logistic regression models. Self-reported rates of having a Pap test in the past three years were higher in women from higher compared with lower SES groups. Compared with women with a bachelor or higher degree, the odds of reporting having a Pap test in the past three years in women with no post-school qualification was 0.86 (p<0.0005). Women with a gross annual income of less than $20,000 had significantly lower odds (OR=0.79) compared with women earning $40,000 or more. Blue collar (OR=0.84) and not employed (OR=0.73) women also had significantly lower odds compared to the referent white collar group. This study reveals differentials in Pap screening behaviour by individual measures of SES in Australia. Area-based SES measures under-estimated the SES differentials in Pap test rates compared with individual measures. Derived population attributable fractions reveal that about a quarter of self-reported under-screening is accounted for by low SES when measured individually, compared to 8% when SES is measured ecologically.

Why is job security lower for Maori and pacific island workers? The role of employer‐provided training

Maori and Pacific Island workers are more likely than other workers to transit from employment to unemployment whereas exit rates from unemployment are relatively less dispersed across ethnic groups. Hence, lack of job security appears to raise the unemployment rate for Maori and Pacific Islanders. This paper uses unit record data to examine one source of lower job security, which is that Maori and Pacific Island workers may receive less employer‐provided training. The results suggest that training reduces employer‐induced separations, but even controlling for training, and for age, years of schooling, gender, marital status and industry, there is an identifiable effect of Maori and especially Pacific Islanders experiencing higher involuntary job losses. Moreover, once the available characteristics of workers and their jobs are controlled for, Maori have the same incidence of training and may receive even more intensive training than do Pakeha workers. In contrast, Pacific Island and Other ethnic group workers have a significantly lower level of employer‐provided training. Thus, job training does not explain all of the ethnic differences in involuntary unemployment rates and some ethnic minorities appear to receive a lower level of job training which cannot be explained by their known social characteristics and their occupation and industry. Of course there may be social characteristics for which we have not been able to control, which further explain part of the differences in experiences between the ethnic groups.

Cervical screening in migrants to Australia.

To examine differentials and time trends in self-reported Pap test rates by migrant status from the 1989/90 and 1995 Australian National Health Surveys (NHS). Unit record data for females with the variables of interest were extracted from the 1989/90 and 1995 NHS and combined. The dichotomous outcome variables were 'ever had a Pap test' and 'had a Pap test within three years'. The principal study factor was country-of-birth, but language spoken at home (English or not) was also examined. The indirect age-standardised screening ratio was used to calculate proportions of 'ever had a Pap test' and 'had a Pap test within three years' and differences were tested statistically using logistic regression analysis for each year of survey by migrant status. Odds ratios for rates of reporting 'ever had a Pap test' were significantly lower in women born in southern Europe, Italy, other countries, southern Asia, Middle East, Greece and South-East Asia compared with Australian-born. Reported rates of 'ever had a Pap test' were significantly higher in the 1995 NHS (p<0.001). There were significant increases in screening for the Australian-born, New Zealand-born, and women born in southern Europe, South-East Asia, South Asia and Italy, and both English and non-English speakers over the 1989/90 and 1995 NHSs. Odds ratios for reporting 'had a Pap test within three years' showed significantly lower ORs for women born in the UK, Other countries, Middle East, Greece, and South-East Asia compared with the Australian-born. This study reveals differentials in reported Pap test behaviour by country-of-birth in Australia and that reported screening rates have improved from the 1989/90 NHS to 1995 NHS in most country-of-birth groups.

A Comparison of Income and Expenditure Inequality Estimates: The Australian Evidence, 1975–76 to 1993–94

Using Australian unit record data this paper compares income and expenditure inequalities over the period 1975–76 to 1993–94. The study finds inconsistencies between the two inequality movements over much of this period. We also observe differences in the nature of income and consumption disparities. Both approaches show that the ‘within group’ inequality dominates the ‘between group’ component when the population is divided into household types. The inequality estimates are sensitive to the equivalence scale used as the household size deflator but not to the cost of living index used as the price deflator.

Capital Gains, Homeownership and Economic Inequality

The paper examines the distribution of capital gains and the net benefits of homeownership in terms of the debate between Marxists and Weberians on how homeownership modifies class inequality. Capital gains and net benefits are measured using unit record data from valuation records of the 1980s and a survey carried out by the Australian Institute of Family Studies in 1991 of over 2500 families in Sydney and Melbourne. Higher socio-economic groups generally had significantly larger dollar gains than groups below them and, by this measure, homeownership adds to inequalities generated through the labour market. However, because they had commenced purchase with smaller deposits, the low socio-economic groups did as well or better than other groups when capital gains are expressed as a percentage return on deposits. Capital gains are one of the variables included in a measure of net housing benefits; an attempt to quantify all the major benefits and costs involved in purchase rather than renting. The net housing benefits accruing to different socio-economic groups since deregulation of finance markets are compared using survey data relating to the 1980s property boom, and estimates using the much lower rates of interest and capital gains in housing markets of the 1990s. Again, there is a strong association found between dollar benefits and socio-economic status. The paper concludes that, measured in dollar terms, the distribution of net housing benefits in Sydney and Melbourne has had a substantial class bias since deregulation, tending to create a more unequal society.

Estimated alcohol consumption in the 1995 National Health Survey: Some methodological issues

ObjectiveThis paper critically examines the methodology used to ascertain alcohol consumption in the 1995 Australian Bureau of Statistics National Health Survey (NHS). MethodThe 1995 NHS was the second in a series of regular, five-yearly population surveys designed to obtain national benchmark information on a range of health-related issues and enable the monitoring of trends over time. The first survey was conducted in 1989-90.In the 1995 NHS, respondents were asked about the quantity of each of seven types of alcohol consumed on up to three days of the week before the interview day. In the 1989-90 NHS, however, respondents were asked about alcohol consumption on each of the seven days before the interview. This paper uses Unit Record Data from the 1989-90 National Health Survey to investigate the methodology used in 1995. ResultsIt is shown that the estimates of alcohol consumption obtained using the 1995 methodology are highly dependent on the day of the week on which the interviews were conducted. ConclusionIt is recommended that the 1995 methodology not be used in future NHS surveys. The 1989-90 methodology of seven-day retrospective diaries would seem a better choice for future National Health Surveys.

The Economic Progress of Maori Men 1986-1996

Data for the Census years 1986, 1991 and 1996 show that Maori/non-Maori differences in employment and unemployment rates increased over the ten-year period. In search for a possible explanation, multinomial log it models are estimated with unit record data in order to control for changes in socio-economic and demographic factors such as age, qualification, and family and parental situation. The analysis shows that the Maori/non-Maori differences in employment (unemployment) that are explained by the multinomial logit models decreased from 51% (43%) in 1986 to 38% (34%) in 1996. Two potential contributors for declining relative labour market outcomes of Maori men are identified: increasing returns to skills and changes in the sectorial composition of the workforce.

The Wage Curve and Local Labour Markets in New Zealand

Blanchflower and Oswald argue in their 1994 book that there is a stable downward-sloping convex curve linking the level of pay to the local unemployment rate. They derived this so-called wage curve from measurements on individuals within regions (local labour markets) for several countries and periods. Other investigators have confirmed the robustness of this finding. In this paper we seek evidence for the wage curve in New Zealand drawing on data at the regional level by means of the /996 census of population and dwellings. New Zealand research is hampered by the inaccessibility of unit record data and the paper reports results based on publicly available grouped data. The results show that a cross-sectional wage curve does exist in New Zealand. The elasticity is in the range of-0.07 to -0.12, which is similar to results obtained for other countries. However, research to date has not been able to choose between competing explanations for this phenomenon. We argue that a better understanding of the dynamics of local labour markets is an essential requirement for further study of the wage curve.

Comparison of Alternative Models of Household Equivalence Scales: The Australian Evidence on Unit Record Data*

The equivalence scale is an important parameter in view of its use in welfare comparisons. This paper investigates sensitivity of the scale to models, methods and commodities. The estimations range from single equation. OLS in case of Engel and Rothbarth models to maximum likelihood in case of demographic demand systems. The estimated scales from demand systems generally lie between the Engel and Rothbarth scales. The Equivalence Scale Exactness hypothesis, considered necessary for interpretation of the scale as child cost, suffers only a mild rejection. The Australian evidence suggests that rank two demand models, rejected by the data, may exaggerate child costs.

Poverty and health: exploring the links between financial stress and emotional stress in Australia.

The links between poverty and health have been investigated in a number of independent studies, as well as by the Poverty Commission in the 1970s and, more recently, the National Health Strategy. However, much of the poverty research suffers from the lack of detailed information on health status, while the work conducted in the public health sphere has used rather rudimentary poverty measures. The research reported here attempts to overcome these limitations by using unit record data from two national household surveys conducted by the Australian Bureau of Statistics in 1990. Together, these two data sets contain an enormous amount of detailed information on household incomes and the health status of individuals. Data from the two surveys are combined in a way which allows the links between poverty and health to be explored in greater detail than has hitherto been possible in Australia. Analysis of the integrated data set focuses on the links between poverty and several measures of stress-related poor health. The results from a variety of different methods point to the existence of significant differences between the reported incidence of stress of those whose incomes place them either side of a poverty threshold. The size of the statistical association between poverty and stress is of both numerical and statistical significance, although further work, preferably using longitudinal data, is needed on the important issues of causation.

Census Unit Record Data

This article reviews Census record file releases (1981, 1986, 1991) and surveys the extensive literature based on these data sets.

CRIME AND FEAR: Evidence from Australia

This study, which is based on unit record data from the Queensland Crime Victim Survey of 1991, conducts, using a common set of explanatory variables, a joint analysis of the probabilities of becoming, and of being afraid of becoming, a victim of personal and of housing crime. It then proceeds to analyse the relationship between previous victimization experience and fear of crime (personal and housing). Its broad conclusion is that unlike personal crime, where reducing the fear of crime--over and above concern with reducing the incidence of crime--might be an important policy objective, the policy goal for housing crime should, primarily, be to reduce its incidence. This is because, for personal crime, variations in the probability of being afraid greatly exceeded variations in the risk of becoming a victim: fear of personal crime thus emerges as a problem in its own right. On the other hand, for housing crime, incidence is an accurate reflection of fear. It also concludes that lack of neighbourhood cohesion, neighbourhood incivility and perception of relatively high neighbourhood crime levels contribute significantly to the probability of being afraid of crime and to the risk of victimization. It suggests that, in policy terms, community action might be a more effective means of combating both crime and the fear of crime than a leave it to the police, that's what they get paid for' attitude.

Income, Health and Happiness

Abstract This article utilises unit record data from the 1989–90 National Health Survey to explore the relationships between low income and perceptions of health and happiness. After describing how the key variables were constructed, attention is focused on the incidence of perceived poor health and unhappiness among those individuals living in income units on either side of a poverty threshold related to the Henderson poverty line. The results reveal that those in poverty generally have poorer health and are less happy than those above the poverty threshold. The observed differences are statistically significant for many income unit types, particularly non‐aged single people and couples with children. Attention then turns to exploring how health and happiness vary across the deciles of the income distribution. Again, significant differences are revealed, particularly when more sophisticated distributional measures are employed. Finally, consideration is given to the possibility that the results reflect reverse causality, and that the observed correlations incorporate the role of intervening variables.