2011 HILDA Survey Research Conference: Abstracts

Thursday, 14 July 2011 - 11:00 am
SESSION 2A

Are Women Really That Happy at Work? Australian Evidence on the `Contented Female’
Temesgen Kifle and Parvinder Kler
This paper investigates the apparent paradox of females possessing higher levels of job satisfaction compared to their male counterparts despite having (on average) less secure jobs, lower promotion opportunities and receiving lower take home pay. Using panel of HILDA Waves 1-8, we test the four main theories that attempt to explain the aforementioned genderjob satisfaction paradox; these being (a) differing gender expectations, (b) sample selection bias, (c) gender-bifurcated personal and employment characteristics as well as (d) genderdivergent work values.

A parsimonious sweep of the statistical analysis finds in favour of the paradox, thus justifying a more in-depth investigation. Postulating that female (but not male) entry into the workforce is heterogenous by age, education and the presence of children, this paper divides the genders into four groups; the aggregated, educated, young and childless as well as young (irrespective of parental status). As well, we expand our study to encapsulate not just overall job satisfaction but also five other aspects of job satisfaction satisfaction with pay, satisfaction with job security, satisfaction with the type of work undertaken, satisfaction with the hours worked and satisfaction with work-life balance.

The paper finds evidence of significant gender differences in employment characteristics in all four groups, and while gender differences are less apparent with regard to personal characteristics, they do however appear in both the employed-only as well as overall samples.
As well, sample selection bias does not appear to be present in our study. Further statistical analysis also indicates that across the six measure of satisfaction, females do report higher rates of satisfaction relative to males across the four groups, though this is weaker for the educated group. As well, changes over time are observed for the educated and young (childless) groups. Ordered probit results that control for expectations and characteristics paint a muddier picture however. There are no longer any gender differences in satisfaction across all six measures for the young (childless), while for the educated group, females are ess satisfied with two measures with another two being statistically insignificant. The paradox is still somewhat apparent for the young (irrespective of parental status), while only exhibiting its presence strongly in the case of the aggregated group, though even then not uniformly so.

In sum, the paper finds that the paradox does exhibit itself when employed females are not disaggregated by age, education and parental status. The paradox weakens and even disappears when such factors are incorporated into the analysis. This suggests that employed females should not be viewed as a monolithic block in the labour force and may also intimate that the ‘lower expectations’ hypothesis for females may not extinguish over time as they become even more significant players in the employment market; instead, expectations may change among females depending on their education and parental status.

 

Gender differences in rates of job dismissal: Why are men more likely to lose their jobs?
Roger Wilkins and Mark Wooden

Several studies of job loss have noted a higher rate of job loss among male employees than among female employees, but there has been almost no investigation of why this might be the case. In this paper, we draw on Waves 1-9 of the HILDA Survey data to consider the roles of alternative potential explanations. In common with findings of studies drawing on other data sources, the HILDA data show that men have a considerably higher rate of job loss, with the proportion of employees dismissed or made redundant each year 40% higher for male employees than for female employees. We hypothesise that the higher rate of job loss among men could be the result of: differences between men and women in the characteristics of those who choose to participate in the labour market; differences in choices of jobs; differences in employee behaviour in employment; and/or differences in employer treatment of men and women (i.e., discrimination against men). Given the information available in the HILDA data, our focus is primarily on the first two potential explanations. Pooled panel probit models of the probability of job loss in the next year are estimated on male and female employees as a function of demographic and job characteristics. Estimates indicate the higher rate of job loss for men is partly a function of differences in demographic characteristics, partly a function of differences in the types of jobs men and women do, and to a limited degree due to differences in behaviour in employment and/or differences in employer treatment of men and women who are otherwise the same. Decompositions that relax the pooling assumption, allowing effects of demographic and job characteristics to differ for men and women, reveal that demographic characteristics in fact account for almost none of the difference in job loss rates. Job characteristics are still found to be important, suggesting men and women do choose jobs with different job loss probabilities. Nonetheless, a substantial proportion of the difference remains unexplained (that is, due to differences in effects associated with given characteristics or factors). We therefore conclude that differences in behaviour in employment and/or differences in employer treatment of men and women are an important source of the higher rate of job loss for men.

 

Thursday, 14 July 2011 - 11:00 am
SESSION 2B

Asking for help from welfare organisations: the evidence from HILDA, 2001 to 2009


Bruce Bradbury and Tony Eardley

A question in several national population surveys, including the HILDA, asks whether respondents have sought help from welfare or community organisations as a result of a shortage of money. This provides as close a proxy as is available of applications for emergency relief from non-governmental welfare organisations in situations of financial crisis or stress.
This paper presents an analysis of data from the nine waves of the HILDA between 2001 and 2009. It focuses on three sets of questions:

The analysis was carried out as part of a wider study of the operations of FaHCSIAs Financial Management Program, which funds emergency relief (ER), financial counselling, money management services and microfinance schemes, aimed at helping people to overcome financial adversity, manage their money, participate in their communities and plan for the medium to long term.

We find that those asking for help from welfare services are likely to have low long-term incomes and high rates of long-term welfare receipt even when controlling for other current indicators of well-being such as income. There is also a strong association between the prevailing unemployment rate and the proportion of the population seeking help. Thus ER acts as a useful adjunct to other measures designed to direct assistance to the most disadvantaged. However, while the program is reasonably well targeted in this respect, it also provides some assistance to people who may be in short-term financial crisis but are not in the lowest income group for long periods of time.


Importance of income, participation and social support for the wellbeing of people with disability
Peng Yu
Money can be among the most intuitive yardsticks for understanding the value of things, tangible or not. Following the practice of Powdthavee (2008), who has put a price tag on friends, relatives and neighbours in the UK, this paper demonstrates the relative importance of participation and social support compared to income for the subjective wellbeing of people in Australia using a shadow pricing method.

Based on the first nine waves of the HILDA data and a fixed-effects approach, this paper reveals significantly positive relationships of economic and social participation, social support and income with self-reported overall life satisfaction. Further, it highlights the relative importance of participation and social support in relation to income, especially among people with work restrictions due to disability as illustrated below.

Using a shadow pricing method, in terms of affecting life satisfaction, moving from full-time employment to unemployment gives a loss in wellbeing equivalent to up to $165,000 household disposable income per annum (adjusted with the modified OECD equivalence scale and measured in 2001 dollars) for people without disability, and for people with work restrictions due to disability the corresponding income loss equivalent to such a move doubles ($333,000). An active club membership is associated with the equivalent in wellbeing of $38,000 equivalised household income per annum for people without disability and $92,000 for people with work restrictions due to disability. A one-unit increase in perceived social support (on a scale from 1 to 7 with the sample mean of 5.5) is associated with $59,000 and $79,000 equivalised household income per annum for people without disability and people with work-limited disability, respectively. These estimated effects are not only statistically significant, but also provides insight to the great personal value people put on participation and social support considering the sample mean of equivalised household disposable income per annum being less than $34,000. In contrast, the effect of income on life satisfaction is not significantly larger for people with disability than for the general population. The implications of the results are discussed in a policy context.

 

 

Thursday, 14 July 2011 - 11:00 am
SESSION 2C

Persistence of Social Exclusion among Older People in Australia: What are the protecting
Riyana Miranti and Peng Yu

The existing literature on social exclusion among older people, though relatively limited, suggests that disadvantage among the elderly is cumulative in nature. Some aspects of disadvantage starting at early life stages have long-term consequences. As such, older people with disadvantages may be subject to higher risks of persistent social exclusion. This paper aims to improve understanding of the persistence of social exclusion among senior Australians in three ways. First, incidence of social exclusion among older people using selected indicators is analysed. Second, it examines whether an older person experiencing social exclusion at one time is more likely to experience it again (persistence). Third, it investigates what factors may be protecting older people from persistent social exclusion.

The analysis is conducted using the first eight waves of the Household, Income and Labour Dynamics in Australia (HILDA) Survey (20012008). The sample of the elderly is disaggregated into a younger group (55-64 years) and an older group (65+ years). Notwithstanding a low incidence of social exclusion (especially in multiple dimensions), the analysis indicates a relatively high persistence of exclusion among those who experienced social exclusion in the previous year, about 45 per cent experienced the same situation in the following year. Further, results from panel data econometrics suggest that higher education and income, as well as better health conditions and previous employment experiences are important protective factors from social exclusion for older people.

 

The quality of the place of living, social participation and well-being in older age: evidence from HILDA
Wojtek Tomaszewski

This paper explores the links between the quality of the place in which older people live, their social participation, and their well-being, using data from the Household, Income and Labour Dynamics in Australia (HILDA) survey. Recent changes to life style observed in developed societies, including Australia, coupled with advances in medical sciences and increased availability and improved quality of health care, have resulted in people living longer and healthier lives than ever before. However, increased life expectancy and structural population ageing also potentially places more individuals at risk of social isolation and social exclusion. Understanding processes behind social exclusion in older age, including barriers to social participation and achieving personal well-being, is paramount to ensuring that social policies are in place to help older people stay independent and prevent unnecessary institutionalisation and dependence (Willis and Dalziel, 2009).

Much research on social exclusion has focused on understanding the links between living conditions, social participation and individual well-being, and the importance of the quality of the place of living for outcomes in older age has been firmly recognized (Burridge and Ormandy, 1993; Stirling 1997; Ashmore, 1998; Marsh et al., 2000; Matte & Jacobs, 2000; Hood, 2005; Donald, 2009), although research has most often focused on the effects on physical health, rather than social participation or subjective well-being. Aiming at filling this gap in the literature, this paper hypothesises that environmental conditions, captured in terms of housing and neighbourhood characteristics, can have independent and potentially cumulative effects on older people's opportunities for playing their part in society and, ultimately, achieving personal well-being and independence.

The main independent variable is the quality of the place of living, described at two levels: local area and individual housing. Both aspects are captured by indicators included in HILDA, such as the external conditions of dwelling, the quality of the neighbourhood, overcrowding, and the perceived adequacy of the individual accommodation. The paper investigates their impact on the domains of social participation (captured in HILDA by the set of questions on social interactions and membership in clubs/community participation) and subjective well-being (captured by HILDA’s measure of life satisfaction).

The research exploits the longitudinal aspect of the data by including in analyses up to nine waves of the survey (depending on the specific model). The selected sample covers all older people (aged 65+) living in private households and inactive in the labour market, including those living with younger panel members. One of the hypotheses tested in the paper is how the presence of other people in the household impacts on social participation and the quality of life of older people. The paper also tests the interaction between the quality of individual accommodation and the local area: whether older people living in poor housing have better outcomes if they live in better neighbourhoods, or conversely they are more excluded because they feel they do not fit.
Random effect models are used to estimate the ‘net effect’ of the place of living on social participation and well being, while controlling for other key factors, such as health status, income, gender, family composition, rurality and tenure.


Thursday, 14 July 2011 - 1:40 pm
SESSION 3A

Labour market engagement of mature-age workers
Jennifer Poehl

As previous research indicates, there are many factors that influence labour market engagement of those approaching retirement. This paper examines those factors with a particular focus on the role of employment and job characteristics. The analysis centres on the labour market transition of men and women aged between 45 to 64 years old over a (approximately) one year time period. The data used for our analysis is drawn from the first 7 waves of the HILDA data and we utilise a multinomial logit (MNL) model to determine the characteristics associated with remaining in full-time employment, moving into full retirement and/or moving into partial retirement. In addition, we are conducting an array of robustness checks including a fixed effects model.

Consistent with previous research, we find that certain demographic characteristics such as age, health status and marital status as well as some financial factors such as wages, government benefits and house ownership influence the retirement paths of mature-age workers. Further the results indicate that certain job characteristics significantly influence the employment engagement of older workers. The factors with a significant influence include union membership, contract type and firm size.

 

The dynamics of satisfaction with working hours
Robert Breunig and Xiaodong Gong

Surveys such as HILDA find a substantial fraction of workers who express dissatisfaction with their current working hours. Recent claims in the Australian media and pronouncements by government officials, drawing upon these figures, postulate that these dis-satisfied workers provide a large pool of under-utilised labour. But to what degree is this true? In this paper, we show that dis-satisfaction with working hours is by-and-large a transitory state. Using the dynamic panel nature of HILDA, we show that the vast majority of workers who feel that they are working too little actually increase their hours in subsequent waves and express satisfaction with their working hours in future waves. Thus, dis-satisfaction seems to be a temporary state through which people pass on the way to finding the job that suits them. This would seem to contradict the view that there exists a permanent stock of dissatisfied workers who can magically serve as a pool of untapped labour supply. We also show that workers dis-satisfaction with working hours due to working too many hours is quite different and is much more persistent.

We explore the relationship between this persistence and household financial stress. Furthermore, we explore the transitions between satisfaction and dis-satisfaction with working hours and their relationships with employment, unemployment, and retirement.

 

Thursday, 14 July 2011 - 1:40 pm
SESSION 3B

The Effect of Own Life Events on Own Mental Health
Cindy Mervin and Paul Frijters
In this paper, we use seven years of data (Waves 2 to 8 for the period 2002-2008) from the Household, Income and Labour Dynamics in Australia survey to estimate the effect of nine life-events on mental health for individuals aged 15 and over. Our analysis has three focal points: whether individuals adapt to life events, the one-off income required to compensate individuals for experiencing a life event, and the investigation of the effects of measures of social support, with a particular focus on marital status, kids, friends, and social network. To investigate these issues we use fixed effect models. There is no adaptation to having a serious illness and being a victim of violence. As a result, the monetary compensations required for constant utility are higher for these events compared to other events where adaptation is complete. Being married significantly buffers against the adverse effect of having a serious illness (e.g. reduces it by 12 percent) and being a victim of violence (e.g.

 

Understanding Housing Affordability as a Determinant of Health: Isolating a Mental Health Effect of a Change in Housing Affordability


Kate Mason

Evidence about the mental health consequences of unaffordable housing is limited. Using data from seven waves of the Household, Income and Labour Dynamics in Australia (HILDA) survey we investigated whether people whose housing became unaffordable experienced a corresponding deterioration in their mental health (measured using SF-36 MCS), over and above other forms of financial stress. We hypothesised that measurable mental health effects of poor housing affordability would be limited to lower income households as high housing costs will reduce their capacity to purchase other essential non-housing needs (e.g. food). Importantly, we sought to isolate the effect of unaffordable housing from the effects of living in a low income household.

We used fixed-effects longitudinal regression to analyse 38,610 responses of 10,047 individuals aged 25 to 64 years who participated in the HILDA survey between 2001 and 2007. We compared within-person changes in mental health in two groups of respondents: those who remained in affordable housing over two consecutive waves (reference group) and those who had moved from affordable to unaffordable housing over two waves (comparison group). For individuals living in low-to-moderate income households, entering unaffordable housing was independently associated with a small corresponding decrease in their mental health score independent of changes in equivalised household income, or having moved house (?=-1.19, 95% CI: -1.97, -0.41). As anticipated, we found no evidence of a similar association in higher income households. We conclude that entering unaffordable housing is detrimental to the mental health of individuals residing in low-to-moderate income households.



Thursday, 14 July 2011 - 1:40 pm
SESSION 3C

Parental Divorce and Trust
Tarja K. Viitanen

This paper examines the effect of parental divorce during childhood on generalised trust later on in life using Australian HILDA panel data for 2005-6. The dependent variable is composed from variations in agreement with the statement: “Generally speaking, most people can be trusted”. The main explanatory variables include the occurrence of parental divorce for the whole sample of 15-29 year olds and the age at which parents divorced for the sub-sample. The analysis is conducted using pooled OLS, pooled ordered probit, and an instrumental variables regression to account for possible endogeneity issues. The results are not significant for men for whom parental divorce does not have a significant effect on their expressed level of generalised trust. However, the results for women indicate that their level of generalised trust is significantly affected by parental divorce with women who have experienced parental divorce expressing lower levels of generalised trust.

 


Thursday, 14 July 2011 - 3.40 pm
SESSION 4A

The Role of Unobserved Heterogeneity and On-the-Job Training in the Employer Size-Wage Effect in Australia
Lixin Cai and Jeff Waddoups

The positive relationship between employer size and wages is a ubiquitous feature of advanced industrialised economies. The purpose of the present study is to clarify the nature of the employer size-wage effect in Australia by determining the extent to which it can be explained by observed and unobserved quality differences, including difference in on-the-job training. The empirical results are based on analysis of the Household Income and Labour Dynamics in Australia (HILDA) Survey, which is a nationally representative panel data set focused on family income, employment, and well-being. Our findings indicate that for males, quality adjusted employer size-wage effects are quite small and mostly driven by lower wages for workers in the smallest firms (fewer than twenty workers). For females, size-wage effects disappear when unobserved quality differences are accounted for. We also find that accounting for differences in the incidence of job training has no effect on the structure of wage

 

Job Mismatches and Labour Market Outcomes: Panel Evidence on University Graduates
Kostas Mavromaras, Seamus McGuinness, Nigel O'Leary, Peter Sloane and Zhang Wei

The interpretation of graduate mismatch manifested either as overeducation or as overskilling remains problematical. This paper uses annual panel information on both educational and skills mismatches uniquely found in the HILDA survey to analyse the relationship of both types of mismatch with pay, job satisfaction and job mobility. We find that overeducation and overskilling are distinct phenomena and that their combination results in the most severe negative labour market outcomes. Using panel methodology reduces strongly the size of many relevant coefficients, questioning previous cross-section results and suggesting the presence of considerable unobserved heterogeneity which varies by gender. The results also suggest that it is on overskilling and particularly its combination with overeducation that policy should focus.

 

 

Thursday, 14 July 2011 - 3:40 pm
SESSION 4B

Life Satisfaction and Retirement: A Latent Growth Mixture Modeling Approach
Lachlan Heybroek

This research investigates changes in life satisfaction across the retirement transition and examines both pre- and post-retirement phases. Retirement is an important life course event, potentially involving major changes to an individual’s economic and social circumstances. Some individuals may enjoy increased time to pursue leisure and other activities after retirement, while others may find retirement a period of uncertainty, isolation and financial hardship. It is hoped that this research will contribute to a better understanding of how retirement is experienced in Australian society by identifying groups of people who experience retirement differently and examining variations in the characteristics and resources of those groups. The research has potential implications for social policies designed to improve the welfare and wellbeing of older Australians.

Latent Growth Mixture Modeling (GMM) was used to investigate individual trajectories in life satisfaction across the retirement transition. The analytical sample is based on the first 9 waves of HILDA and includes 600 people (302 men, 298 women) who made a single transition into retirement and who stay retired. A model including discontinuities three years before retirement, at retirement and three years after retirement provided a better fit to the life satisfaction data, than linear, quadratic and two component piecewise models.

Three groups of respondents experiencing different trajectories of life satisfaction after retirement were identified by the models. The first group, by far the largest comprising 88% of respondents, showed relatively stable life satisfaction and reported the highest overall levels of life satisfaction both before and after retirement. A second group, comprising 7% of respondents, experienced a decline in life satisfaction in the three years leading up to retirement, followed by an increase after retirement. For a third and final group, 5% of respondents, life satisfaction steadily declined in the three years after retirement. Covariates were included to model life satisfaction trajectories within and between the three latent groups. Within the three groups, higher expected life satisfaction at retirement was associated with being older at retirement, being in a relationship, having high levels of social support and having better health. To distinguish between the three groups, similar covariates, in particular being healthier, were identified as determining the high and stable levels of life satisfaction observed for the majority of respondents in group 1.

Overall the results show that while most people experience very little retirement related change in life satisfaction, there are latent groups who experience significant change. Those people who do experience significant change tend to have lower levels of life satisfaction associated with worse health and less access to a range of social and economic resources. These findings indicate the benefit of identifying different groups with varying trajectories as opposed to simply examining overall trends.


Has retirement made Australians healthier and happier? Evidence from HILDA surveys
Ya-Seng (Arthur) Hsueh

As life expectancy continues to prolong, financing retirement can be as challenging as financing health care for any aging society. As a result, governments are considering prolong the age of compulsory retirement. However, regardless of the financial implications for the government and individuals, whether retirement (versus continuing working) is actually good for health or overall well-being of aged people is another important question to ask.

Previous research has conflicting findings regarding this research question. Some findings suggest that working is good for both physical and mental health as well as overall well-being for aged people, while other findings indicate the opposite. This research has two objectives. One is to add further evidence to this research question. The other is to overcome the limits of previous research by using 9 waves (years) of nation-wide representative longitudinal panel survey data from Australia Household, Income and Labour Dynamics in Australia (HILDA) and refined econometrics models.

In this research, all Australians aged 45 and above in the first wave (year 2001) will be included for analysis and followed up for 8 years. The 2009 wave is the most recent HILDA data available to the public. Physical health and mental health are measured by widely used health status measurement instrument SF-36. Happiness is measured by multiple questions. Control variables include a wide range of comprehensive information on personal attributes, health behaviour, social economic status, social capital and condition of living environment.

Methodologically, the potential problem of endogeneity of health and retirement can be reduced by two approaches. One is to use the 2-stage random effect model by choosing appropriate instrumental variable. The other is to use the fixed effect model on people who were in good health (both physically and mentally) before and at the year of retirement. The results from the two approaches will be compared and discussed. The impact of retirement on health and happiness/wellbeing will be examined by controlling for confounding factors such as aging, wealth, education, health behaviour, social activities and support, and quality of living environment. The findings of this research can shed the light on important public policy debates on whether and how exactly prolong compulsory retirement age will impact on aged Australians’ health and overall wellbeing/happiness.


Thursday, 14 July 2011 - 3:40 pm
SESSION 4C

Relationship transitions and womens time on domestic labour in Australia: A multiprocess, multilevel approach
Michele Haynes, Janeen Baxter, Belinda Hewitt and Mark Western

In this paper we examine data from 8 waves of the Households, Income and Labour Dynamics in Australia (HILDA) survey to investigate the impact of marital status transitions on domestic labour time. Although there is a growing body of literature on this topic, previous research has failed to adequately address selection issues or the role of unobserved factors in determining transitions in marital status and time on housework. Neglecting these factors potentially becomes more consequential with declining rates of marriage, increased rates of cohabitation, remarriage and non-partnering. Specifically, the unobserved factors influencing the transition into or out of marriage may be the same as those influencing time spent on housework. The new multiprocess, multilevel models presented here enable examination of these issues. Our aim is to first use HILDA data to identify the joint influence of unobserved factors on marital status transitions and women’s time on domestic labour and second, to determine whether there is a selection effect into and out of relationships for women who do varying amount of domestic work.

A simultaneous-equations model is used to jointly examine the relationships between time on domestic labour and marital status transitions to allow for correlation between unobserved partner and person characteristics that impact on each process. Our results show that women who spend more time on housework when single, suggesting higher levels of domesticity, spend more time on housework after cohabitation or marriage. But there is no evidence of selection of these women into marriage. We also found no evidence to support the hypothesis that women who do varying amounts of housework are more likely to separate from cohabitation or marriage.

Overall we conclude that the unobserved factors influencing time spent on housework are not related to the unobserved factors influencing marital status transitions.

 

Fluctuations in the unemployment rate and intra-household allocation of time
Susmita Roy

Shifts in the unemployment rate affect individuals through both actual and potential loss of a job. This paper contributes to the literature by exploring whether fluctuations in the unemployment rate induce couples to readjust time allotted for market work and housework and hence leisure time. More specifically, I enquire whether a rise in spouse's and in one's own sectoral unemployment rate affects one's weekly time allocation. A rise in the unemployment rate as opposed to actual job loss influences one's time allocation through a reassessment of one's job loss probability. An increase in the odds of a job loss reduces one's bargaining power within the household and this might lead to a fall in one's leisure time relative to his/her spouse. This is the bargaining effect. On the other hand, in response to a rise in partner's job loss probability, the other spouse might increase his/her hours of market work, resulting in the reduction of his relative leisure time. This is the efficiency effect. This study utilizes the first seven waves of HILDA survey to explore which of the two effects dominate.

To identify the coefficient of interest, this study constructs a series of unemployment rate proxy that varies across sector-state-year. Next, I tie this measure to individuals based on their primary sector of employment. Preliminary evidence shows that couples respond to a rise in the unemployment rate by reducing their hours of housework without a concomitant increase in their hours of market work. This is true of both husbands and wives with the exception of women and men cohabiting with a partner. Women in a de-facto marriage respond to a hike in the unemployment rate by increasing their hours of market work and cutting back on housework. Analogously, men in a de-facto marriage respond to a rise in the spousal unemployment rate by increasing their hours of market work and cutting back on housework.

The results have important implications for household welfare given the public good nature of housework. All components of housework might not be marketable and if this is true then periods of high unemployment might result in a fall in the welfare of even those households, where both the husband and the wife have a job.

 

 

 


Friday, 15 July 2011 - 8:50 am
SESSION 5A

Mobility and Mobility Contexts: Modeling and Interpreting Residential Change in Australia
William A.V. Clark

Residential mobility and migration are the processes whereby families change their houses and their residential locations whether it is a neighborhood, a city or a state. Mobility and migration have been central in modern western society especially with the economic expansion of cities in the 20th century and have been fundamental in changing metropolitan neighborhoods and communities. A well-established body of research has documented that residential mobility is a function of age, tenure, family status, the demand for living space and changes in household composition. What is less well understood are the effects of place on residence change and the outcomes of residential change, especially the changing levels of satisfaction as an outcome of residence change. The research in the current paper uses the most recent waves of the HILDA data (waves 8 and 9) to examine these two questions neighborhood effects and mobility outcomes.

Models of mobility confirm the applicability of the disequilibrium approach to residential change in Australia. Households who are younger move more often and family change precipitates residential change and for longer distance moves being unemployed increases the probability of moving. The disruption of divorce and separation, as expected, increased the probability of moving. However, housing space as measured as persons per bedroom does not show a significant relationship with the probability of moving. This is contrary to findings in Europe and North America and may reflect the design of the variable. The findings largely parallel similar studies in North America and Europe and a much earlier study in Australia (Hassan, et al, 1996) although that study did not incorporate measures of family change or community contexts.

There is considerable interest in evaluating the role of neighborhood and place more generally in both mobility and satisfaction (Shields and Wooden, 2003). Introducing measures of home, neighborhood and community satisfaction provides support for the relevance of place effects on mobility. The measures of neighborhood and community satisfaction are significant and I conclude that geography and place matter in the likelihood of moving.

Place seems to matter more for local moves than long distance moves but satisfaction with neighborhood as well as the house decreased the likelihood of a long distance move.

While the models are useful interpretations of mobility likelihoods, a decomposition of reasons for move by distance raises questions about the classic housing short distance/ employment long distance explanation for residential relocation. Three decades ago we reliably argued that households moved long distances for employment reasons and short distances to adjust their housing needs (the Rossi explanation) and both were linked to categories of the life cycle. Now with changing family structures, the increase in two worker households and changing organization of labor markets, that dichotomy is less persuasive as a conceptual structure for understanding residential change. Clearly, the models still privilege housing tenure for local moves and employment for long distance moves but not only is housing an important factor in long distance moves, family reasons and other events complicate the former simple dichotomy. The decomposition of the detailed data on moves in the HILDA survey shows that family and life style and unintended moves all play a role in the complex decisions to change locations. Nearly forty percent of those who move more than 30k (interpreted in this paper as those who change labor markets) cite family reasons as the basis of their change in location and only one third overall say that they moved for employment reasons.

The long term interest in residential change comes from the argument that for the most part “we move to improve”. The question which then arises is whether a household is more satisfied after moving and how does that satisfaction relate to the reason for moving. That is do those who move for housing, jobs, or neighborhood reasons have greater or lesser satisfaction related to their reason for moving? Overall, movers are more satisfied after their move. The outcomes vary with respect to house, neighborhood, community and jobs. While more than half report an increase in satisfaction with house and neighborhood only 38 % report being more satisfied with job opportunity outcomes. Ordered logit models of the change in satisfaction suggest that overall the move does improve but it does not necessarily improve access to employment.

What is new and how do the findings increase our understanding of mobility in Australian contexts? The rich detail of the HILDA survey is unusual in the depth of questioning on both mobility and levels of satisfaction. The data allow an intensive examination of the disequilibrium models of mobility and the role of local contexts in the mobility decision. The examination of reasons for moves and their translation into residential outcomes paints a positive picture for housing change but a less positive scenario for job opportunity satisfaction. The examination of reasons for moves reiterates the finding that mobility behavior is more complex than a dichotomy of local housing moves and long distance employment moves. That the models have modest explanatory levels reiterates the complexity of moves for Australian households.

 

Socio-spatial impacts on labour market transitional outcomes
Sharon Wheller

This paper will examine how location affects both the duration of unemployment and employment transitional outcomes in Australia with a particular focus on those living in regional urban localities outside major cities.

The study uses an unbalanced and unweighted panel from HILDA waves 1 - 8 and includes observations from 17,258 individuals. A total of 60,743 single episodes are derived from the employment diary to provide more dynamic information on transitions between labour market states throughout the survey period. The research explores aspects of both the conventional definition of unemployment as well as the broader concept of underemployment that includes discouraged job seekers or those marginally attached to the labour market and those working part-time but wanting to work more hours.

Building on the spatial labour market literature, a four level model has been developed to examine what social and spatial factors contribute to the length of time spent in unemployment in regional communities. Dependent variables have been grouped within the following broad categories: individual or human capital level effects; social reproduction of the labour market or family level effects; social embeddedness or community level effects; and state and regulatory, or national level effects.

Findings suggest that the length of time spent in unemployment is influenced by a variety of factors of a spatial and social nature, but the factors contributing to other forms of employment disadvantage did not necessarily operate under the same dynamics. Transition out of one 'poor' employment episode was often not the end of employment disadvantage, particularly for those living in regional towns.


 

Friday, 15 July 2011 - 8:50 am
SESSION 5B

Relationship Transitions and Subjective Wellbeing: A Longitudinal Analysis
Janeen Baxter and Belinda Hewitt

We examine trends in subjective wellbeing across marital status using 9 waves of HILDA data. We advance previous research by examining two measures of wellbeing happiness and life satisfaction, examining a wide range of possible marital statuses and examining variations within couples. Our analyses differentiate those who are single and not in a relationship, those who are in a relationship but not living together and those who are in a relationship with a live-in partner, either cohabiting or married. We compare results for cohabiters who plan to marry from those who do not, and examine differences between those in a first marriage and those in a higher order marriage, as well as respondents who are separated, divorced and widowed. We estimate a series of fixed effect models on each of the outcome variables that control for unmeasured heterogeneity and also hold constant key independent variables likely to influence wellbeing. Results indicate that men and women who are married have higher levels of wellbeing than those who are not married. We find that transitions into relationships, marriage or cohabitation, significantly increase wellbeing while transitions out of relationships because of separation, or widowhood, negatively impact on wellbeing. We find no gender differences in these patterns and no significant differences between cohabitation and marriage.

 

Does Homeownership Improve Personal Wellbeing?
Steve Stillman and Yun Lian

Does homeownership help people become better citizens, parents and neighbours? Advocates for promoting homeownership claim: “ownership is consistently linked to improved life outcomes for individuals as well as to more cohesive communities” (Skilling 2004). Benefits may include better health and educational outcomes, greater security and independence, and more political and social participation (New Zealand Government 2006; Skilling 2005; DiPasquale and Glaeser 1999). International studies have found that higher levels of homeownership are associated with a variety of positive outcomes (for example, Rossi and Weber 1996; Rohe and Stegman 1994).

However, there is conflicting evidence on whether homeownership has a positive impact on mental and physical health and there is evidence that homeownership can be damaging for groups with declining health and those with an uncertain ability to make mortgage payments (Nettleton and Burrows 1998; 2000).

This paper uses longitudinal data from nine waves of the nationally representative Household Income and Labour Dynamics of Australia (HILDA) survey to examine the causal relationship between homeownership and a number of measures of personal wellbeing. In particular, we examine the impact of being a homeowner on life satisfaction, financial satisfaction, satisfaction with family relationships and mental health. As HILDA follows all original sample members regardless of whether they have moved to other residences in Australia and collects detailed information about their current housing circumstances, we are able to estimate fixed effects regression models that allow us to control for unobservable differences in individuals that are related to both homeownership and personal wellbeing. This modelling approach allows us to potentially identify the causal relationship between homeownership and personal wellbeing.

Our main analysis focuses on longitudinal samples of households that have maintained the same composition (‘single’ versus ‘couple’) over multiple waves of HILDA while potentially changing homeownership status. This allows us to isolate the impact of homeownership from the impact of changes in relationship status, which are often correlated. For these samples, we estimate three specifications of fixed effects models; a first that controls for no other covariates, a second which controls for household composition at a disaggregated level (e.g. the number of adult, children and the average ages of both adult and children), and a third which also controls for adult labour force status and total household income, which are potentially correlated with changes in homeownership status and personal wellbeing.

We then extend this analysis along a number of dimensions. First, we examine whether changes in homeownership have differential effects on individuals who also change couple status at the same time. Second, we examine the impact of time in a particular dwelling on personal wellbeing. Third, we examine whether the relationship between homeownership and wellbeing varies with the amount of housing debt. Fourth, we examine whether there is an interaction between effect size and personality. Fifth, we quantify the extent to which any measured effects operate through intermediate variables, such as changes in marital status, number of children, labour force participation, and household income.


Friday, 15 July 2011 - 8:50 am
SESSION 5C

Women Rule: Preferences and Fertility in Australian Households
Elliott Fan and Pushkar Maitra

Using a unique panel data set from Australia, we investigate how individual fertility preferences translate into fertility realizations. We examine whether the dominance over fertility decisions is held by the wife, the husband, or whoever with higher (or lower) fertility desire. We find consistent evidence that the wife's preference is more important than the husband's preference in predicting subsequent births, no matter whether her initial fertility desire is higher or lower than that of her partner. We also explore the implications of the introduction of so called Baby Bonus on fertility outcomes. The results suggest that the effect of Baby Bonus is heterogeneous - it has no effect on birth decisions for couples sharing a similar level of fertility desire, but it appears to increase the influence of whoever has a higher level of fertility desire for couples whose desires differ significantly.

 

Differences in health and health inequalities between Australia and New Zealand
Fiona Imlach Gunasekara, Kristie Carter and Sarah McKenzie

The Short Form 36 Questionnaire (SF-36) is a widely used measure of health-related quality of life in health and social science surveys and measures various aspects of physical and mental health in adults. We compare SF-36 scores (asked of all respondents aged 15 and older) using data from wave 8 (2008) of the Household Income and Labour Dynamics in Australia (HILDA) Survey and wave 7 (2008/2009) of the Survey of Family Income and Employment (SoFIE), the equivalent New Zealand longitudinal survey. Overall, SoFIE respondents score significantly higher (i.e. better) on both mental and general health SF-36 components than do HILDA respondents. This may be due to differences in the sample designs or administration of the questionnaires. However, the distributions of the mental and general health scores are similar, and vary by age and sex. In both SoFIE and HILDA, men report declining general health with increasing age and highest mental health at older age but young women report poorer general and mental health than young men in women, general health declines over time and mental health improves around middle age. We will also compare concentration curves to examine the distribution of reported health across income and occupation in working age (18-65 year old) men and women, in order to compare health inequalities between Australia and New Zealand.


 

Friday, 15 July 2011 - 10:50 am
SESSION 6A

Estimating Net Child Care Price Elasticity of Partnered Women with Preschool Children Using Discrete Structural Labour Supply-Child Care Model
Xiaodong Gong and Bob Breunig

The purposes of paper are twofold. First, we construct and estimate a joint discrete structural model of labour supply and child care demand for partnered women with preschool children using the ‘in-confidence’ version of HILDA. As a methodological innovation, we impose a quantity constraint that the number of total child care hours (formal plus informal child care) is at least as large as the labour supply of the mother. Second, using simulations, we estimate the gross and net child care price elasticities for all mothers with preschool children and for various demographic sub-groups to shed light on labour supply behaviour and child care demand of women across income levels and socio-economic characteristics.

 

Monetary and Non-monetary Determinants of Nurses' Labour Supply
Barbara Hanel, Guyonne Kalb and Tony Scott

Australia faces a severe and increasing shortage in nurses’ labour supply (Simoens et al., 2005). In order to address this issue, it is important to understand the factors that are driving nurses’ labour supply decisions. Earlier research finds very low wage elasticities among nurses (Benham, 1971; Phillips, 1995; Chiha and Link 2003).

Later research thus paid more attention to other, non-monetary factors that may determine nurses’ labour supply (Shields and Ward, 2001). There are two important characteristics that distinguish the nursing profession from many other professions: first, there is a very high incidence of shift work that might impede the balancing of work and family life, which may be of particular importance in this highly female-dominated profession. And second, the job often implies a high level of emotional and personal involvement.

These specific job characteristics suggest that the labour supply of nursing qualification holders might depend, although not exclusively, highly on their personality and their family circumstances.

HILDA is ideally suited to address this question as it provides not only information on respondents’ children and partner characteristics but also information on the individuals’ personality (collected in Wave 5). I will use the 9 waves of HILDA (2001-2009) to investigate to what extent nurses’ labour supply is driven by five personality dimensions (extroversion, emotional stability, openness to experience, agreeableness and conscientiousness)and by the presence of children and a partner as well as partners’ employment status and income.

The analysis is conducted in two steps. In the first step, multinomial panel models are estimated to explore separately a) if nursing qualification holders choose to work as a nurse full-time, part-time or not at all, and b) which type of shifts are chosen by working nurses. The models will include socioeconomic control variables and the abovementioned personality variables and family variables. At this step, the project addresses the question to what extent nurses’ labour supply depends on these factors directly. The second step uses a structural approach to model the labour supply decision, as established by Van Soest, 1995 for the investigation of female labour supply. The utility function underlying the labour supply decision has hours of work (in different shift-types) and income from work as its arguments. In contrast to earlier research that focused on average wage elasticities among nurses, the parameters of the utility function will be allowed to differ across individuals with different personalities and different family circumstances. At this second step, the analysis addresses the question to what


 

Friday, 15 July 2011 - 10:50 am
SESSION 6B

Time trends in psychosocial working conditions in a representative sample of working Australians 2000-2008: evidence of narrowing disparities?
Lauren Krnjacki, Tony Lamontagne, Anne Kavanagh and R. Bentley

Background: Psychosocial working conditions are important, modifiable determinants of physical and mental health. Acknowledging previous research demonstrating cross-sectional disparities in psychosocial working conditions, we assessed time trends in job control and security in a representative sample of working Australians to determine whether disparities were narrowing. Cross-sectional and longitudinal disparities were assessed by sex, age, occupational skill level, and employment arrangement.

Methods: Measures of job control (5 items) and job security (2 items) were collected in 8 annual waves (2000-2008) from the HLDA survey. Population-weighted measures of control and security were calculated for the whole population, and separately by sex, age, occupational skill level and employment arrangement. Model-predicted time trends were generated using population-averaged longitudinal linear regression models, with year fitted as a categorical variable, and adjustment for potential confounders of sex, age, education and indigenous status.

Differences in time trends by sex, age group, skill level and employment arrangement were tested as interactions with time.

Results: Significant cross-sectional disparities were observed by sex, age, occupational skill level, and employment arrangement. Job control remained relatively flat over time, whereas job security increased from 2000-2007, followed by a decrease at the onset of the global financial crisis. There was no evidence of narrowing of disparities over time, with the exception of an improvement over time in job control among young workers compared to older groups.

Conclusion: Most cross-sectional disparities in job control and security persisted over time, with the positive exception of improving job control among young workers. Concerted policy and practice intervention could reduce persisting inequalities in psychosocial working conditions and associated illness burdens.

 

Understanding Life Satisfaction and Education Puzzle in Australia: A profile from HILDA Wave 9
Cathy Gong, Rebecca Cassell and Marcia Keegan

It is well established from both human capital theory and empirical evidence that higher education attainment can enhance a person’s future outcomes, including bringing about better career opportunities, income and wealth attainment and health outcomes. One would therefore expect people who achieve higher levels of education to be happier on average than those with lower levels of education. However, our preliminary analysis using HILDA 2008 data found that in Australia people with higher education level are less likely to report feeling either “very satisfied” or “not satisfied” with their lives and more likely to report being “satisfied”, indicating that higher levels of educational attainment are not necessarily always associated with higher life satisfaction. Using both the pooled cross-sectional and panel data drawn from the 2001 to 2009 HILDA surveys, we will be able to examine this puzzle and its underlying stories.

HILDA survey data provides us with very detailed information about different aspects of job, life and relationship satisfaction, as well as education attainment, labour market status, individual characteristics, family background and life events. These variables will allow us to fully investigate and understand the complicated relationships between life satisfaction and education achievement, such as whether the well educated people have different satisfaction level on different life aspects when compared to those with low or median education level. Ordered Logit model will be used to estimate the partial effect as well as the causality of education achievement on job and life satisfaction by controlling for other factors, such as income, wealth, marriage, cultural background, number of children and labour market status etc. In order to solve the endogeneity issues and estimate causality,

 


Friday, 15 July 2011 - 10:50 am
SESSION 6C

Using a Life Course Approach to Understand Involvement in Volunteering in Australia
Edith Gray, Siew-Ean Khoo and Anna Reimondos

This paper is based on Stage 1 of a research project funded by the Department of Families, Housing, Community Services and Indigenous Affairs. Stage 1 focuses on the analysis of HILDA data to examine volunteering over the life course. Stage 2 examines volunteering for different types of organizations using Negotiating the Life Course data. We thank the staff of the Research and Analysis branch of FaHCSIA for their comments on the paper.

The data used for this research come from the Households Income and Labour Dynamics in Australia (HILDA) survey funded by the Commonwealth Department of Families, Housing, Community Services and Indigenous Affairs (FaHCSIA) and conducted by the Melbourne Institute for Applied Economic and Social Research (MIAESR). The research findings are the product of the researchers and the views expressed do not necessarily represent the views of MIAESR or the Minister for Families, Housing, Community Services and Indigenous Affairs and cannot be taken in any way as expressions of Government policy.

 

Who provides care? An event history analysis of the effect of sex and relationship status on the provision of informal care in Australia
Alice Falkiner

Informal care provided by family members is the most common form of care for people with disabilities or long term health problems in Australia. The ageing of the Australian population is likely to increase demand for care, and increase the number of Australians providing informal care. It is therefore important to understand who is likely to provide informal care, and at what age Australians are most ‘at risk’ of providing care for family members. This paper assesses the different patterns of providing care by men and women, and by relationship status. As a longitudinal survey, the Household, Income and Labour Dynamics in Australia (HILDA) survey provides an opportunity to examine patterns of care over time. Event history analysis is used on data from waves 5-9 (years 2005-2009) of the HILDA survey. A proportional hazard model (Cox regression), controlling for other socio-economic factors, shows that women are significantly more likely to provide informal care than men and that people living with a partner are more likely to provide care than singles. Survival analysis is used to provide the hazard and survivor functions: the hazard function illustrates the conditional probabilities that men, women, singles and those living with a partner are at risk of providing care at each age, and the survivor function provides the proportion not providing care at each age. Findings from the survival analysis indicate that there are statistically significant differences in survival time between all combinations of sex and relationship status, except between single men and single women.


 

Friday, 15 July 2011 - 1:30 pm
SESSION 7A

Interrelationships and Social Determinants of Residential Location Moves and Unemployment
Duration in Australia
Angela Higginson, Michele Haynes, Will Probert and Paul Boreham

In this paper we investigate the interrelationships between transitions from unemployment to employment, the duration in unemployment and changes in residential location. We examine the social determinants of these processes and whether unemployment duration is influenced by previous episodes of unemployment and residential location moves. The economic and social costs associated with unproductive time spent in labour market transitions between jobs and between unemployment and employment in Australia are high. A large component of these costs is the financial and social cost borne by individuals and communities, some of which are absorbed by welfare payments. However, many families with adults in unemployment are not able to absorb these costs and an exit from unemployment to employment may only be possible with a change in residential location. The geographical space in which people reside and work has implications for their ability to move between jobs, particularly for those living in regional and remote locations. Identifying the social and regional determinants of those individuals at risk of long-term unemployment can inform social policy to help minimise the duration of unemployment. Using a life-course approach and the first eight waves of the Household, Income and Labour Dynamics in Australia (HILDA) Survey we jointly model two processes: transitions between employment states and residential location moves. Examining the joint dynamics of recurrent episodes of employment transitions and housing moves will enable an assessment of whether individuals who move between residential locations when unemployed are more likely to experience shorter duration in unemployment. Our dataset follows 10,300 participants, aged between 18 and 65 at their entry to the survey, across a maximum of 268 time periods. The dataset is based on the HILDA employment calendar reporting schedule which collects information at three time points each month, matched to the dates of life transitions reported in the annual survey, and annual wave-based covariates. Applying a joint, multilevel multistate discrete time event history model across recurrent episodes of competing employment states and location moves, we identify the characteristics of those most at risk for periods of long-term unemployment. In particular, we focus on whether relocating to a new location to find work increases the probability of successful transition out of unemployment.

 

Family migration and labour market outcomes of partnered women
Anastasia Sartbayeva

This paper uses data from the first nine waves of the HILDA Survey1 to study the effects of family relocation within Australia on married women's labour market outcomes. It uses matched couple data for men and women who were partnered before and after
relocation, and compares labour force participation and earnings in couples who recently moved long-distance to those of the couples who have not moved or moved only a short distance. The descriptive analysis shows that married men and women in the mover and nonmover families have similar employment rates and earnings before relocation, however in the year following relocation the employment and earnings of women in the migrant families are significantly lower than those of non-migrant wives, while the earnings and employment rates of the husbands do not differ substantially. At the same time, wives in the relocating families
report no significant changes in satisfaction with employment opportunities post-relocation, whereas the husbands' satisfaction with their employment opportunities significantly increases for those who have recently moved compared to non-movers.

The two-stage regression models are then estimated to analyse the determinants of long distance migration in couples, as well as the effects of migration on labour market outcomes, accounting for self-selection. The husband's education and employment are found to be stronger predictors of long-distance migration than the characteristics of the wife; however, families where the wife is more highly educated are found to be more likely to move. Finally, the estimation results of employment and earnings models are consistent with the findings of descriptive analysis, confirming that migrant wives have much lower employment rates and
earnings compared to non-migrant married women, unlike the married men for whom the adverse effects of migration are minimal in comparison.

 

 

 

Friday, 15 July 2011 - 1:30 pm
SESSION 7B

Retirement Decisions
Zhou Peng

Using Household, Income and Labour Dynamics in Australia (HILDA) Survey on a panel Probit model, we investigate the gender effects on the retirement decision. We found that the factors affecting the retirement decision with respect to gender are less pronounced than the finding of the previous studies that emphasis on the financial incentive for male and the family reasons for female. The retirement decisions are interdependent on the trade off between household retirement wealth and individual retirement age. Both man and woman show less responsiveness to the Social security benefit.

 

Subjective Wellbeing in Retirement: Evidence from HILDA
Garry Barrett and Milica Kecmanovic

Does retirement represent a state of relatively prosperity or is it a time of unanticipated economic hardship? To address this question we analyse measures of subjective wellbeing (SWB) in the HILDA Survey. Specifically, the research examines individual’s self-reported financial prosperity and standard of living, and how these measures change during the transition to retirement. This empirical research is based, in part, on the HILDA retirement modules (conducted in 2003 and 2007). It is found SWB improves, on average, as individuals enter retirement. However, the average increase in SWB with retirement masks significant disparities across individuals. In particular, individuals who retired due to a major health event or work stress indicate marked declines across all SWB measures considered. Further, individuals who experience a deterioration in their relationships with family and friends, report a substantial decline in SWB with retirement. The research also makes a methodological contribution by using the responses by the set of individuals who retire during the observation period of the HILDA Survey to evaluate the reliability of the retrospective measures of changes in SWB (recorded in the retirement modules) against the contemporaneous reports of SWB recorded during the transition to retirement.


 

Friday, 15 July 2011 - 1:30 pm
SESSION 7C

Modelling wage dynamics among Australian workers
Timothy Higgins

Accurate modelling of earnings dynamics is critical for projections of poverty dynamics and inequality, taxation revenue and savings, and public policy costings. Despite extensive analysis of earnings dynamics using the PSID, there has been little research using Australian longitudinal sources. This paper seeks to partially address this gap by using the first seven waves of HILDA to estimate and compare dynamic wage models. Residuals from a mean fit to hourly wage are decomposed into permanent and transitory components, and it is found that the variance of the components differ by employment state. A random walk for the permanent component can be justified on economic and empirical grounds, and the models are extended to allow for non-Gaussian errors and serial correlation in the transitory component. Short-term predictive ability of the models is mixed, and possible improvements are discussed. The implications of the models developed are illustrated through projections of outstanding debt and repayments through the Higher Education Contribution Scheme for a hypothetical population.


 

 

HILDA Survey Research Conference 2011

Conference Enquiries:

Email: [email protected]
Phone: +61 3 8344 2100