Number 12: How do income support recipients engage with the labour market?

This report was published by the former Department of Families, Community Services (FaCS).

Executive summary

Aims

This report  utilises the Department of Family and Community Services (FaCS) Longitudinal Data Set (LDS) 1% Sample  to examine how  income support recipients interact with  the labour market. The study  covers  the years  1995  to 1999.

The LDS 1% Sample  is a unique data set because it contains detailed longitudinal information on income support recipients over an extended period  of time. The LDS 1% Sample  is based  on Centrelink’s administrative records of income support recipients. Fortnightly records contain information on the personal characteristics and payment details  of pensioners and beneficiaries.

Structure

The paper  begins  with  an overview of the structure of income tests facing  pensioners and beneficiaries and how  these  tests impact  on the labour  market  incentives for income support recipients.

It then explores the take-up of part-time  employment options  by income support recipients. The paper  presents time-series estimates of the incidence of earned income received by pensioners and beneficiaries and develops a statistical model  of the factors  influencing employment take-up.

The LDS 1% Sample  provides a unique insight  into the labour  market  activity of income support recipients because it tracks  those  recipients at fortnightly intervals when  they  are on the Centrelink register. (As an administrative data set, the LDS does not provide  any information on the recipients when  they  move off the Centrelink register.) We utilise  this feature  of the LDS 1% Sample  to provide  summary job and income support histories of recipients. This analysis allows us to investigate some important questions in the dynamics of the labour  market  activity of recipients, which earlier literature has not addressed. Do income support recipients remain  in jobs (while on income support) for long periods or short periods?  Do they  churn  between jobs while on income support? Do they  churn  between periods on income support and off income support?

The final and most important topic  this paper  addresses concerns the role that employment take-up plays  in improving the chances of final exit  from the income support system  for existing income support recipients. The opening up of short-hour  employment options  for recipients provides a potentially important stepping stone  to full-time employment (for those whose preferences are for longer  hours). The question we address  is whether or not the evidence from the LDS 1% Sample  supports the stepping-stone hypothesis.

Focal groups

Five categories of income support recipients are examined in this report:

  1. the young  unemployed, designated throughout the paper  as Unemployed (<21);
  2. the prime-age unemployed—Unemployed (21+);
  3. older  unemployed—Unemployed (60+);
  4. those  income support recipients whose partners are on a pension or allowance and who have no recent workforce experience—Partner Allowance;
  5. sole parents—Parenting Single; and
  6. members of a couple with  dependent children receiving Parenting Payments—Parenting Partnered.

Income tests and labour market incentives

Under the Australian  income support system, pensioners and beneficiaries may work  and receive income from paid employment without automatically losing  their  income support entitlement. Income  support recipients continue to receive income support payments at the maximum rate of entitlement up to a specified level  of earned income (the  so-called  disregard). The disregard is higher  for pensioners than allowees. The rate of withdrawal of income support on receipt of earnings beyond  the disregard is also lower for pensioners as compared with  allowees.

The structure of the income tests applying to allowees (the  unemployed and Parenting Partnered income support recipients in our study)  as compared to pensioners (Parenting Single recipients) has important labour  market  implications. Entitlement to income support is maintained for a greater range  of earnings and at higher  levels  for Parenting Single recipients than for allowees. At the same time, the labour  supply preference structure for Parenting Single recipients may be more skewed towards increased hours devoted to non-work  activities (such as child-care responsibilities). This means  that Parenting Single  recipients are more likely  to optimally locate  themselves at labour  supply positions consistent with  the maintenance of their income support payment. Voluntary  part-time  employment while on benefit  is more likely  to characterise the Parenting Single  recipients group  than the unemployed and, to a lesser extent, ParentingPartneredrecipients. We would, therefore, expect higher  rates of take-up of employment positions together with  a more stable  employment profile  among  Parenting Single income support recipients than other  groups  in this study.

Among the young  to prime  age unemployed income support recipients, lower levels  of voluntary part-time  employment while on income support can be expected. When such recipients obtain  employment while on income support, they  are less likely  to view  that job as an optimal-hours position. Many would  prefer  full-time work. These recipients would, therefore, be more likely  to continue the search  for full-time employment and exit  the income support system  altogether when  they  have found it. Alternatively, young  to prime  age unemployed recipients may continue the full-time job search  process longer  rather  than accept a less-than- optimal  hours offer. Hence, we are more likely  to observe exits  from income support that do not involve  a part-time  employment transitory state among  this group  than among  the ParentingSingle and Parenting Partnered recipient groups.

These observations lead us to expect a somewhat lower take-up of employment positions together with  a less stable  employment profile  for the young  to prime  age unemployed income support recipients than that evident for Parenting Single and Parenting Partnered recipients. Exit rates (the  rate of leaving  income support) should  also be higher  for recipients who  are young  to prime  age unemployed than for those  recipients in the Parenting Single and Parenting Partnered categories.

The stepping-stone hypothesis

Part-time employment positions may not be the preferred employment location for some income support recipients. They would  prefer  a full-time job. Nevertheless, they  may accept part-time  employment offers, in which case they  would  be described as being  involuntarily part-time  employed. Why might  such recipients follow  such a course  of action?  One important reason  for this is that the short-hours  job may increase the chances of the recipient obtaining increased hours of work  as compared with  searching full-time for a job with  the preferred number  of hours. More efficient signalling, better  information flows, on-the-job learning, training opportunities, morale, a reduction in cash flow difficulties, and positive health  effects  may all contribute to this outcome.

As such, higher  rates of exit  from income support for those  already in employment as compared with  those  not in employment may result. Part-time employment works  for these  recipients because it acts as a stepping stone  to full-time employment. In so doing, part-time  employment also leads  to higher  rates of exit  from the income support system.

Employment rate trends

The employment rate, in this study, is defined  as the proportion of income support recipients who  take up employment positions while on income support. As hypothesised, the evidence from the LDS 1% Sample  indicates that the employment rate is indeed highest among  Parenting Single recipients. The employment rate for female  Parenting Single  recipients was around 26 per cent  in 1995  but had risen  to 29 per cent  in June 1999. As a point of comparison, the employment rate among  female  Unemployed (21+) recipients, was not only lower but had also fallen over the period  covered by the LDS 1% Sample; from 24 per cent  in 1995  to around  to 22 per cent  at the end of the period. A similar  decline in the employment rate, but from a lower initial  base, is evident among  male Unemployed (21+) recipients. Among Unemployed (<21) recipients, the decline in the proportion of recipients in paid employment was even  more pronounced over the 1995  to 1999  period. The take-up of employment positions among older  unemployed recipients has always been  very low and remains an obvious  concern for policy makers.

Modelling of the take-up of employment positions while on income support indicates that Parenting Single income support recipients have a higher  probability of being  employed while on income support than do other  recipients, having  controlled for other  determinants of earned income receipt. These determinants include: the number  of dependent children in the case of female  recipients (negative effect), age (positive, but diminishing effect), Aboriginal  and Torres Strait Islander  status (negative), non-English speaking background (negative), government rental status (negative), and job search  activity under  the Activity  Test (very  strongly positive).

Earnings

Our analysis also revealed that the earnings of Parenting Single income support recipients are generally higher  than those  for other  income support payment groups; a natural  outcome given the greater earnings range  available under  the pensions income test relative to the allowance income test. Research studies  have consistently pointed to the higher  incidence of poverty among  sole parents in the Australian  community. Relatively high take-up of part-time employment positions and a strong  earnings profile  helps  to significantly ameliorate both the incidence and severity of poverty among  this group.

One further  result  that is of interest is the gender earnings gap in favour of women in the LDS 1% Sample. This earnings gap may reflect  an hours effect rather  than a wage  rate effect (we have no information on employment hours in the data). Nevertheless, the result  is of some interest in light of the well-documented earnings disadvantage felt in the labour  market  by women generally.

Employment and income support histories

The distinction between the labour  market  experience of unemployed income support recipients and Parenting Single  recipients is put into sharper relief  when  we examine their income support spell  and employment history  profiles. These profiles  are obtained by tracking individual recipients through the various  fortnightly records that make up the LDS 1% Sample dataset. In this analysis, an income support spell  refers  to a continuous period  on income support. An employment episode is defined  as a continuous period  of employment while the income support recipient is on income support. (It should  be apparent that the definition of an employment episode precludes periods of employment when  the individual is not receiving income support payments.)

Our employment and income support histories analysis suggests that the Unemployed (<21) and Unemployed (21+) recipient groups  typically exhibit both shorter  income support spells and shorter  employment episodes than can be found for each  of the remaining income support categories. Moreover, as the income support spell  lengthens, the gap between the mean  duration of employment episodes for Parenting Single  recipients and those  for the Unemployed (<21) and Unemployed (21+) recipient groups  increases. The female  Parenting Single recipient, who  obtains  work  while on income support, spends, on average, twice as much time in employment as a female Unemployed (21+) recipient who  obtains employment while on income support.

Likewise, there  is a tendency for the Unemployed (<21) and Unemployed (21+) recipient groups  to experience a somewhat higher  degree of ‘job churning’ (or multiple employment episodes) than can be found among  Parenting Single recipients. In other  words, Unemployed (<21) and Unemployed (21+) recipients are more likely  than their Parenting Single counterparts to move between different  jobs while on income support. In short, the Parenting Single  pattern of engagement with  the labour  market  is much  more one of long-duration spells of income support and stable  patterns of income receipt than is evident for the two main unemployed recipient groups.

One point of concern for policy-makers in our results  is the relatively low incidence of take-up of part-time  employment positions among  income support recipients with  medium to long- duration spells  of income support. To take one example, among  male Unemployed (21+) recipients with  income support spells  lasting  more than two years, 46 per cent  of recipients had not experienced a single  employment episode during  their  spell  on income support. This is of significant policy concern as the longer  the unemployed person  remains away  from any sort of employment, the more their  skills and for some, their  mental  health  will  deteriorate. An important focus of any employment-related strategy for the long-term  unemployed, given this evidence, is the provision of wage  subsidies to enable many of these  recipients to engage in the formal labour  market.

Our results  also show  that Unemployed (<21) and Unemployed (21+) income support recipients typically experienced two or three  income support spells  over the 1995  to 1999 five-year  period. At the same time, the modal employment episode outcome was that of no employment episodes in these  multiple spells  on income support.

Labour earnings and exit from the income support system

The most important policy issue  we address  in this paper  is whether obtaining work  while on income support improves the chances of final exit  from income support. If so, then the evidence supports the thesis  that part-time  employment provides a potentially important stepping-stone to a job with  sufficiently high wages so that the individual can exit  the income support system.

We define  the exit  rate  as the proportion of income support recipients at time period  t who  are not on income support at time period  t + 1.

A key conclusion from our times series  analysis of the transition from income support to non- support is that the exit  rate for those  income support recipients who  are employed at a given point of time is generally above the exit  rate for non-earners. This result  applies at the aggregate level  and for both women and men but is particularly pronounced among  the Unemployed (<21) and Unemployed (21+) categories. The gap in the exit  rate between earners and non- earners for the Unemployed (<21) and Unemployed (21+) recipient groups  is around  6 to 10 per cent  in the case of the short-run exit  rate (exit  from income support within a month). This gap rises to 15 per cent  in the case of the long-run exit  rate (exit  from income support within a year). The gap between earners and non-earners narrows considerably, however, in the case of Parenting Single recipients. This result  is in line with  our previous hypothesis that the voluntary part-time  employment rate is considerably higher  for this category of recipients than for the prime-age unemployed.

The role of part-time  employment in affecting an exit  from the income support system  is further  addressed in the paper  via the use of hazard rate  modelling. Here the focus is not on the proportion of recipients who  exit  the income support system  at a particular calendar time point, but on exit  from income support at fortnightly points  within given  income support spells. The hazard  rate for a given  fortnight extract date within an income support spell  is derived by dividing the number  of recipients in a chosen income support category, who were observed to exit  from income support in that fortnight, by the number  of recipients who potentially could  have exited in that fortnight. Our estimated models  of the hazard  rate confirm the important role for exits  of employment episodes while on income support for the unemployed. Part-time earnings have a significant positive effect on the hazard  rate controlling for all other  effects. Our results, therefore, strongly support the stepping-stone hypothesis. The take-up of employment options  among  income support recipients aids the movement off income support and into longer-hours employment for those  recipients who  prefer  such jobs.

[ Return to Top   Return to Section ]

1 Introduction

In the Australian  income support system, pensioners and beneficiaries may work  and receive income from paid employment without automatically losing  their  income support entitlement. Income  support recipients continue to receive income support payments at the maximum rate of entitlement up to a specified level  of earned income.1 Beyond  the income test free area (commonly referred to as the disregard), additional income reduces the value  of an income support payment. The rate at which an income support payment is reduced as labour  earnings increase depends on a range  of factors, but principally on whether the recipient is covered by the ‘allowance’ income test or the ‘pension’ income test. The latter  test is more generous than the former reflecting a higher  (indexed) disregard and lower income support payment withdrawal rates.

The provision of an earned income supplement supports a number  of social  policy objectives. Because labour  earnings do not reduce income support payments dollar for dollar, the total income of the pensioner and beneficiary rises as labour  earnings increase. As a result, poverty is alleviated and the degree of income inequality reduced. The earned income supplement also promotes the private financing of expenditures and, consequently, self-reliance among  recipients.

From a labour  market  perspective, the structure of the earned income supplement affects the labour  market  incentives facing  the income support recipient. When an income support payment is withdrawn sharply on receipt of labour  earnings, the incentive to engage in the labour  market  is blunted. A moderate taper, on the other  hand, increases the incentive for existing recipients to obtain  paid employment.2 Moreover, the opening up of short-hour employment options  provides a potentially important stepping stone  to full-time employment for those  recipients who  prefer  to be employed for longer  hours but cannot  find suitable longer hours employment.3In this case, the provision of an earned income supplement acts to provide  an effective link to final exit  from income support.

In this paper, we utilise  the Department of Family and Community Services (FaCS) Longitudinal Data Set (LDS) 1 per cent  Sample  to explore the role of earned income in the Australian  income support system, focusing particularly on the role that part-time  employment can play in assisting the movement off income support for pensioners and beneficiaries.4 The LDS 1% Sample  is a unique administrative longitudinal data set drawn  from Centrelink’s fortnightly records of income support recipients. Fortnightly records contain  information on the personal characteristics and payment details  of pensioners and beneficiaries and their  partners. Results presented in the paper  are based  on the LDS 1% Sample  over the period  July 1995  to June 1999.

Four main topic  areas  are covered in this paper:

  1. The channels through which income tests influence the labour market incentives faced by pensioners and beneficiaries. How does the earned income supplement affect optimal labour  supply choices?
  2. Employment rates  and the distribution of earnings. Have employment rates among income support recipients increased or decreased over the period  1995  to 1999. How important are factors  such as income support payment type, age, and family status to the take-up of paid employment positions while on income support?
  3. Employment episode and income support histories. Do income support recipients who  obtain  jobs remain  in continuous employment for long periods of time while on income support or do they  churn  between jobs? Do they  churn  between periods on income support and off income support?
  4. Exit from income support. Does the receipt of labour  earnings while on income support positively influence the rate of exit  from income support?

Two broad income support payment programs are examined in this study. The first is the income security for the unemployed program covering payments such as Newstart Allowance, Youth Allowance, the Mature Age Allowance, and Partner  Allowance. The second  major category is the income security program for families  with  children covering Parenting Payment  for sole parents and couples.5 Together these  programs represent a significant part of the Australian income support system. They also represent the key income support system  programs in regard to the take-up of part-time  employment options  among  income support recipients.

The structure of the paper  is:

Chapter 2 contains essential background material for the study. The growth of part-time  and casual  employment in the Australian  labour  market  is surveyed and an overview is undertaken of the role of an earned income supplement in social  security systems. The chapter also examines the structure of the income test applying to the selected payment categories in Australia  and how  this structure may, in theory, impact  on the labour  market  incentives facing income support recipients.

Chapter 3 briefly  surveys  the LDS 1% Sample, the data set used in this study  to examine the interaction of income support recipients with  the labour  market. This chapter shows  the structure of the data set, key variables listed  and trends  in the frequency of income support recipients in the unemployed and families  with  children payment programs.

Chapter 4 presents time series  estimates, for the 1995  to 1999  period  of the employment rate (that  is, the proportion of pensioners and beneficiaries in employment receiving earned income). It then proceeds to examine the factors  that influence the take-up of employment options  among  income support recipients using  a logit model  of earnings receipt. Chapter 4 discusses the distribution of earnings among  recipients.6This distribution is clearly influenced strongly by the nature  of the income tests applying to different  recipients.

A number  of important policy-related issues  are addressed in Chapter 4. Does the evidence suggest an increasing or decreasing take-up of part-time  employment positions among  income support recipients over the period  of the LDS 1% Sample?  To what  extent are the trends  in paid employment rates among  recipients affected  by labour  market  conditions? To what  extent are they  affected  by the structure of the income support system  itself and reforms  to this structure over time? How important are socio-demographic variables, such as age and family status, in the receipt of earned income while on income support, relative to income support effects  such as income support payment type?  How does the distribution of earned income differ between the various  payment categories? What impact  does the structure of income tests that apply  to different  income support recipients have on this distribution?

Chapter 5 extends the analysis of how  income support recipients engage in the labour  market while on income support by considering the employment episode and income support ‘histories’ of pensioners and beneficiaries. An income support spell  refers  to a continuous period  on income support. An employment episode is defined  as a continuous period  of employment while the income support recipient is on income support. (It should  be immediately apparent that the definition of an employment episode precludes periods of employment when  the individual concerned is not receiving income support payments.)

In this analysis, we track  individual income support recipients over the period  of the LDS 1% Sample. A record  is kept  of the number  of separate times we record  them in the data (or the number  of income support spells  they  experience), the length  of these  spells  and the frequency and duration of periods of part-time  employment while on income support. There is a lack of relevant Australian  research on how  individual recipients engage with  the labour  market  over a given  income support spell  and through time. The reason  for this is the absence, prior to the LDS 1% Sample, of longitudinal data sets containing sufficiently detailed income support payment and employment information on income support recipients. This paper  helps  to fill this crucial gap in our knowledge of the labour  market  experience of recipients.

In the context of our histories analysis, we ask a series  of questions. Do income support recipients who  gain employment in the labour  market  (while on income support) hold long- duration jobs or short-duration jobs? On average, what  proportion of time spent  on income support is accounted for by periods of paid employment among  those  who obtain  jobs? Do recipients experience a high degree or a low degree of job churning—the repeated movement into and out of employment—within a given  income support spell. How is the degree of job churning related to the type  of income support payment a recipient is on; how  is it related to the length  of the income support spell?  Our analysis covers  both the full set of income support spells  in the LDS 1% Sample  as well  as spells  from a particular group  of income support recipients: namely  the 1995  group  of recipients. The benefit  of focusing on the 1995  group of spells  is that we can follow  a given  cohort  over time rather  than reporting results  for a mixture of years.

While  the histories analysis this paper  undertakes provides a unique insight  into the labour market  experiences of income support recipients it does suffer from one important limitation. The LDS 1% Sample  only records information on those  receiving support. As an administrative data set, it does not continue to display information on former recipients. Hence, gaps appear in the histories of individuals in the LDS 1% Sample  for those  periods of non-income support. The implications of this limitation of the LDS 1% Sample  data will  be made  clear  at various  points in the paper.

We complete our histories profile  of income support recipients by presenting the results  from a year-on-year transition analysis. A transition analysis presents information on a set of people at a given  point in time and then summarises their  position at some specified future  time point. Transition  analyses, therefore, display snapshots of the same set of individuals in the LDS 1% Sample  at different  points  in time. All income support recipients at a particular starting  point date are assessed as to their  income support payment category at that point and as to whether they  are in receipt of earned income. Recipients in this cohort  are assessed again  at one-year, two-year and three-year time intervals. At these  checkpoints, their  exit  status is determined. If they  have left the income support system, we ascertain whether or not they  exited while in paid employment. If not, we determine their  income support payment category and earnings status.

Chapter 6 extends the analysis of transitions undertaken in Chapter 5 by focusing on the role played by labour  earnings in affecting the rate of exit  from the income support system. We first present trends  in exit  rates for income support recipients over the 1995  to 1999  time period. The time series  exit  rate is defined  as the proportion of recipients at time period  t who  are not on income support at time period  t+1. The analysis distinguishes between short-run exits  and long-run exits. A short-run exit  is defined  as a period  off income support lasting  more than 28 days. A long-run exit  is an observation at a one-year  interval of an exit  from income support. The important policy question we address  is whether exit  rates are higher  (or lower) for earners as compared with  non-earners.

Our final piece of analysis on exit  from income support involves  the modelling of the so-called hazard  rate. The hazard  rate gives the probability that an individual leaves  a given  state at a particular point in time conditional on being  in that state prior to that point. In the present context, the hazard  rate gives the probability of exiting income support at a particular point in time, given  that the individual had been  on income support up to that point. In Chapter 6 we present empirical hazard  functions for the prime-age unemployed group  (aged  21 and over and below age 55) and for female  income support payment recipients in the single  and partnered Parenting Payment  categories. The empirical hazard  function  plots the hazard  rate at different fortnight points. Cox’s  proportional hazards  model  is then estimated for these  income support categories to examine the role of various  determinants, including the receipt of earned income, in affecting the hazard  rate.

The report  closes  with  a summary statement of the results  and an outline  of the policy implications of our analysis. Our summary statement links up results  from various chapters of the report  to provide  a typology of the way  income support recipients from the unemployed and families  with  children programs interact with  the labour  market.

[ Return to Top   Return to Section ]

2 Background

2.1 Part-time employment in Australia

2.2 Earned income and the social security system

2.3 Income tests applying to Australian income support payments

2.4 Illustrative simulations

2.5 Summary

One important reason  for the interest in the role of labour  earnings in the income support system  is the growth in part-time  and casual  employment in Australia. Income  support payment recipients who  earn income in the labour  market  while on an income support payment do so, generally, from part-time  and casual  jobs. The structure of the income tests that apply  to recipients, means  that a full-time job would  invariably result  in the loss of income support payments. Indeed, since ‘part-time’ employment in Australia  is defined  as employment of less than 35 hours per week, many part-time  positions will  also result  in the loss of benefit  as a result of the application of income tests.7 As explained in the following section, this is particularly the case for recipients covered by the so-called  allowance income test.

Section  2.1 provides a brief overview of the relevant literature relating to part-time  and casual employment in the Australian  labour  market  and its relationship to social  security structures. Our background review then considers, in section 2.2, how  the application of income tests and the taxation system  may affect, in theory, the incentives facing  income support recipients.

Section  2.3 details  how  the income test is applied to relevant income support payment categories in Australia  and how  it interacts with  the taxation system  to determine the Marginal Effective Tax Rate (METR) profile  that faces individual income support recipients. The METR measures the increase in tax that is paid and the loss of income support that results  from the additional dollar of private income. It is the METR profile  that summarises the joint effect of the income test, benefit  entitlement, and the tax system  on the disposable income of the recipient. It, therefore, acts as a summary indicator of the impact  of the tax-transfer system  on the economic incentives facing  income support recipients in terms  of their  labour  supply choices.

2.1 Part-time employment in Australia

There is an extensive literature on the nature  and extent of part-time  employment both in Australia  and internationally. The trend  to increasing part-time  employment and to growing casualisation of the labour  force was first noted  after the recession of 1974–77 (Gregory  & Duncan 1980  and Kaspura  1984). It was a trend  that was by no means  confined to Australia  and one which has gained increasing research attention (Blanchflower & Corry 1986, Blanks 1990, Dawkins  & Norris 1995, Flatau & Simpson  1996, Junor 1998, Leeves 1997, NBEET 1992, Romeyn 1992, Rubery  et al. 1991, Sadler & Aungles  1990, and Wooden  1996).8

As set out in Table 1, the proportion of workers employed on a part-time  basis continues to grow in the Australian  labour  market  in the 1990s. Part-time employment now comprises over one-quarter of total employment. As Flatau and Simpson  (1996) observe, the recent Australian experience has also been  one of a high persistence of part-time  employment, especially of young females. In other  words, those  who  do obtain  part-time  employment remain  in these  jobs for relatively long periods.

Table 1: Australia: part-time employment as a proportion of total employment
Years Proportion
1990–91 21.7%
1991–92 22.9%
1992–93 23.5%
1993–94 23.8%
1994–95 24.4%
1995–96 24.6%
1996–97 25.2%
1997–98 25.6%
1998–99 26.0%

Another feature  of the Australian  part-time  employment experience, which has relevance for the present study, is the relatively high degree of involuntary part-time  employment in the labour market. By involuntary part-time  employment we mean  the preference for more hours of work by those  employed on a part-time  basis (that  is, working fewer  than 35 hours). Around 34 per cent  of all men who  are part-time  employed wish  to work  more hours. This proportion rises to around  50 per cent  for men aged  25 to 34. The involuntary part-time  employment rate is much lower for women than men. Only 21 per cent  of women are involuntarily part-time  employed. (Among married women in the 25 to 34 age group, this proportion falls to 16 per cent.) Put another  way, the majority of women who  are part-time  employed are working their  optimal number  of hours given  the constraints they  face. While  women tend to have considerably lower involuntary part-time  employment rates than do men, they  make up a greater proportion of the overall  part-time  employment market. Women  comprise 72 per cent  of all part-time employed workers.9

The emphasis in the part-time  and casual  employment literature is on establishing the economic forces  determining the observed trends. On the demand  side, these  include wage  and non-wage labour  costs, labour  productivity trends, and sectoral growth patterns. On the supply side, the increased labour  force participation of women, the increasing extent of dual participation in work and education by students, and the impact  of the social  security system  on labour  market incentives have been  emphasised (Bradbury 1999, McCall 1996, Lewis, 1990, and Robertson, 1989). It is the latter  supply-side influence that we are most interested in.

2.2 Earned income and the social security system

There is a voluminous literature on the labour  market  effects  of social  security arrangements (Atkinson  & Micklewright 1991, Barrett & Cragg 1998, Bradbury  1999, Gritz & MaCurdy  1997, Keifer 1988, Kalb 1998, Katz & Meyer 1990, Keifer & Neuman 1989, Lancaster  1990, McCall 1995, 1996, 1997, Meyer 1990, Narendranathan, Nickell & Stern 1985, Narendranathan & Stewart 1993, Stevens  1997, and Trivedi & Baker 1983). As Atkinson  & Micklewright (1991) observe, the existing research supports a view  that a negative relationship exists  between the level  of the unemployment benefit  and the duration of unemployment but that the relevant elasticity is low. However, of equal  and perhaps greater importance to the level  of the benefit  is the structure of the social  security program in question. Little research has been  undertaken on the question of how  income security arrangements relating to part-time  and casual  earnings affect the duration of unemployment  and the persistence of part-time  employment while on income support. An exception is the recent work  of McCall (1996, 1997). McCall tests the hypothesis that increasing the disregard (the  level  of part-time  earnings at which withdrawal of benefit  begins), increases the probability of part-time  and overall  re-employment and decreases the duration of unemployment of unemployment insurance beneficiaries. His findings  using  United States longitudinal data confirm  these  hypotheses.

Poverty traps

While  empirical studies  on the interaction of part-time  employment and income support arrangements remains limited, the analysis of poverty trap outcomes has attracted considerable policy interest (see  Bradbury  1999, Haveman, 1996, OECD 1994, 1997, Whiteford, Bradbury  & Saunders, 1989, and Whynes 1993). When disregard levels  are low or non-existent and benefit withdrawal rates are close  to, or at, 100 per cent, the additional dollar of earned income significantly reduces benefit  income, leaving  total gross income largely unchanged. In these circumstances, a poverty trap is said to exist  in the sense  that there  exists  little  incentive to engage in the labour  market  for those  income support recipients seeking short-hours employment.

We can illustrate this principle in terms  of the standard  static  labour  supply model  (see Killingsworth, 1983).10 Consider  the case of the individual represented in figure  1. In the standard model, the individual determines the optimal  (that  is, utility  maximising) number  of hours they  wish  to work  given  their  preferences over ‘leisure’ (or more generally non-work  time possibilities) and consumption goods, and their  after-tax  wage  rate. In this example, we assume a wage  rate of wi and, for simplicity, a zero tax rate. The horizontal axis specifies the number of hours of non-labour  activity undertaken (n). T represents the maximum number  of hours available and so T–n represents the number  of labour  hours supplied. All other  goods are on the vertical axis; normalising price  to $1 implies that vertical axis values  represent consumption expenditure.

Assume that the individual is eligible for a benefit  of BT that applies with  no disregard and a 100 per cent  benefit  withdrawal rate. The budget  line facing  the individual is YZBT. Each hour of labour  reduces the benefit  at a $1 for $1 rate. In this case, optimal  labour  supply is zero hours of employment (given  the shape  of this individual’s indifference curves) and yz  consumption of all other  goods. The classical poverty trap case applies and there  is no inducement to supply positive hours. Compare  this outcome with  the no-income-support position. As is clear  from figure  1, the optimal  labour  supply is T–n1  in the absence of the benefit.

Figure 1: The effect  of a 100 per cent withdrawal rate on hours of employment

Figure 1: The effect of a 100 per cent withdrawal rate on hours of employment

Source: Authors.

Voluntarily part-time employed income support recipients

Now consider the case where the income test for the benefit  is relaxed so that initially a disregard applies and the withdrawal rate is set at a level  below 100 per cent. Figure 2 represents this scenario. The budget  line now  facing  the individual is YWUBT. As can be observed from figure  2, the optimal  labour  supply point is given  at T–n2. The individual remains on income support but now  has an incentive to supply positive hours of labour. The introduction of a disregard and the reduction in benefit  withdrawal rates, therefore, provides an incentive to search  for work  while on income support.11

In the case presented in figure  2, the individual chooses to take up an employment position while on income support. We can describe this case as one of a voluntarily (part-time) employed income support recipient. Unless there  are significant changes in preferences, wage rates, or income support entitlements, the recipient illustrated in figure  2 would  continue to remain  in both employment and on income support with  no preference to exit  either  the income support system  or their  employment position. Exit from income support is non-optimal (for the recipient) as, by definition, any movement from an optimal  labour  supply position for voluntarily part-time  employed persons represents a loss of welfare, all other  things  being  equal. As a corollary, we would expect, in the case of voluntary part-time  employed income support recipients, persistence in both the income support spell  and in the episode of employment.

Figure 2: The effect  of an earnings disregard and a less than 100 per cent withdrawal rate on hours of employment

 

Source: Authors.

Involuntarily part-time employed income support recipients

To this point we have focused  on the case of the voluntarily part-time  employed income support recipient. However, there  also exists  another  category of recipients who  obtain  short- hour employment positions. This category of recipients is comprised of those  that prefer longer-hour jobs—that  would result  in movement off income support—but face constraints in the labour  market  of one form or another  in obtaining preferred longer-hour positions. They accept short-hour  jobs as the next  best option  and stay on income support, perhaps at reduced entitlement. We refer to this group  of recipients as the involuntarily part-time  employed.

The importance of a part-time  employment position to the involuntarily part-time  employed group  of recipients is that the short-hour  job may increase the chances of the recipient obtaining increased hours of work  as compared with  the alternative of searching full-time for a job with  the preferred number  of hours. There are a number  of reasons  why  this may happen. First, by being  in the job, the recipient may obtain  more information about long-hours  job openings not available on the open  market. Second, employers have a greater chance of learning about the attributes of the worker if that worker is already in an employment position and as a result  may offer the recipient increased hours of work. Third, on-the-job learning and possible training opportunities may enhance the skill base of the recipient, so making  them more ‘employable’. Fourth, the longer  the recipient stays in their  short-hours  job, the better  is the labour  market ‘signal’ they  emit. Finally, as reported in our research elsewhere (Flatau, Galea & Petridis  2000), the longer  the individual remains out of employment the worse their  mental health  status may be. Obtaining  a job reduces the incidence and severity of mental  health problems. Together these  factors  provide  the basis for the stepping-stone analogy. Part-time employment works  as a stepping stone  to full-time employment and clearly also to exit  from the income support system.12

The substitution into income support

For existing income support recipients, the social  security financing load falls with  a relaxation of the income test because some existing recipients supply more labour  hours. This, in turn, reduces their  income support entitlement through the application of the income test. However, as the poverty trap constraint is relaxed, so more individuals who  were formerly not eligible to receive income support and who  are in paid employment become automatically eligible for income support. In addition  to this immediate accounting effect, the relaxation of the poverty trap constraint results  in behavioural disincentive effects  for those  at the margin  of income support. Preferred labour  supply may fall for such individuals, in an analogous manner, to the rise in preferred labour  supply for individuals currently in income support. If take-up of income support is relatively high, the net social  security financing task may rise. The higher  tax payments required to finance the additional social  security burden  may also produce deleterious second  round  employment effects. The above analysis, therefore, suggests that there  are employment take-up, and exit  rate trade-offs surrounding the relaxation of poverty trap constraints.

2.3 Income tests applying to Australian income support payments

We now  turn our attention from the theoretical model  presented in the previous section to that of the Australian  income test structure.

With the exception of sole parent  income support recipients, all recipients examined in this paper  are covered by the so-called ‘allowance’ income test.13 For the period  of the study (1995 to 1999), the allowance income test incorporates a disregard of $60 per fortnight. The withdrawal rate was set at 50 cents  per dollar for earned income between $60 and $140  per fortnight and 70 cents  per dollar for private income over $140  per fortnight. The disregard threshold point and the withdrawal rate bracket limits are not indexed for allowees and are not increased for the number  of dependents.

The income test for both the Partner Allowance and for the Parenting Payment  (Partnered) takes  account of both the partner’s income and the individual’s income. The partner’s income, however, begins  to affect the recipient’s rate of payment only after the partner’s own  income support payment is exhausted. Partner’s income then reduces the income support payment at 70 cents  in the dollar. The recipient’s own  income affects payment at the same rate as for allowees generally.

The pension income test applies a higher  disregard than the allowance income test. At 1 July 2000, the disregard for sole parents receiving the Parenting Payment  with  one child  was $130.60 or more than double  the corresponding level  under  the allowance income test. Each additional child  raised  the disregard by $24.60. The disregard limit for payments covered by the pension income test is indexed in July each  year  in line with  movements in the CPI. A single withdrawal rate applies above the disregard. Prior to 1 July 2000, the withdrawal rate was 50 cents  in the dollar. (From 1 July 2000, the withdrawal rate was reduced to 40 cents  in the dollar, but this change in the income test is outside the relevant time period  for this study).

2.4 Illustrative simulations

Figures  3, 4 and 5 illustrate how  the receipt of earned income is treated under  the Sole Parent Pension  income test as compared with  that for Newstart Allowance recipients. The analysis is conducted for end-1996.14

Figure 3: Gross labour  earnings and income unit  marginal  effective  tax rates (METR), Parenting  Single plus two  kids, end 1996

Figure 3: Gross labour  earnings and income unit  marginal  effective  tax rates (METR), Parenting  Single plus two  kids, end 1996

Source: Authors’ simulation model results.

Figure 4: Gross labour earnings and income unit marginal effective tax rates (METR), Unemployed (21+), end 1996

Figure 4: Gross labour earnings and income unit marginal effective tax rates (METR), Unemployed (21+), end 1996

Source: Authors’ simulation model results.

Figure 5: Gross earnings and income unit marginal effective tax rates (METR), Unemployed (21+) earner plus spouse receiving Parenting Payment, two kids, end 1996

Figure 5: Gross earnings and income unit marginal effective tax rates (METR), Unemployed (21+) earner plus spouse receiving Parenting Payment, two kids, end 1996

Source: Authors’ simulation model results.

For the purposes of illustration, we assume  a two-child sole parent  income unit. Two Newstart Allowance recipient case studies  are examined. The first is a single  Newstart Allowance recipient. The second  is a Newstart Allowance recipient whose partner is eligible to receive the partnered rate of Parenting Payment. In the latter  case, only the Newstart Allowance recipient is assumed  to receive earned income. In each  case, we examine how  the additional dollar of earned income impacts on the sum of income and non-income support payments, the level of disposable income, and the METR facing  the income support recipient. Add-ons in terms  of rent assistance and mobility allowance are not incorporated in the analysis.

Earned income is on the horizontal axis. Disposable income and private income, both assessed on an income unit basis, are drawn  against  the left-hand  side axis (specified in terms  of $ per week (pw)). The METR profile  is to be read against  the right-hand  side scale. The METR measures the increase in tax paid plus  Medicare Levy (net  of rebates) plus the loss in both income support and non-income support payments received for the additional $1 of gross earned income. In our analysis, tax is levied  and payments provided on an individual basis but the effects  are then aggregated to the income unit level.

Sole parents

The sole parent  simulation results  are displayed in figure  3. At zero earned income, the sole parent  income unit received a maximum total payment $284.05 pw in 1996. This is comprised of a Sole Parent Pension  payment (as it then was in 1996) of $173.20, and non-income support payments totalling $110.85 composed of Family Payment  of $93.10 and Guardians  Allowance of $17.75. The latter  two non-income support payments are non-taxable. Earned income up to the disregard leaves  entitlement untouched and simply  adds to the gross income of the income unit. The Sole Parent Pension  is not reduced for additional dollars  of earned income. Beyond  the disregard, ($73  pw end-1996), gross income is increased by 50 cents  in the dollar with  the remaining 50 cents  being  deducted from the Sole Parent Pension. As is evident in figure  3, the slope  of the disposable income line becomes flatter  as a consequence of the reduction in the income support payment. The METR is equal  to 50 per cent  at this point.

At $240  private income per week, the sole parent  income unit begins  to pay a net tax amount. The METR increases as a consequence from 50 per cent  to 66 per cent. The rise in the METR from 50 per cent  to 66 per cent  reflects not only the effect of gross tax payments but also the impact  of the loss of rebates. The kink in the budget  line facing  the income unit arises  from a tax effect and not a benefit  effect. This emphasises the point that the incentive structure facing the income unit is affected  by both tax and transfer  effects. The disposable income line flattens out. Notice that the METR for the sole parent  income unit rises again  from 66 per cent  to 75 per cent  at an earned income level  of $377  pw as a result  of the move into the 34 per cent marginal tax rate bracket.15

Entitlement to the Sole Parent Pension  continues up to $419  pw of earned income. Interestingly, 10 per cent  of female  adult full-time employees received weekly earnings of less than $419  pw in 1996. This suggests that the very low paid may receive some entitlement to income support even  if they  are in full-time jobs.16 At $419  pw, the sole parent  income unit moves off the Sole Parent Pension, but retains  entitlement to Family Payment. At $450  pw the Family Payment  itself would  be reduced and the METR would  increase again  (this effect is not illustrated in our figure). The contour of the METR line now  reflects the effect of gross tax payments and loss of rebate.

The ‘passport’ for concessions is also lost at $419  pw. At the point of passport loss, the non-cash income loss may be relatively significant for some income support recipients. METRs derived solely  in terms  of cash components, underestimate the degree of the total resource loss at the point of passport loss.

The single unemployed

Consider  now  the case of the single  Newstart Allowance recipient. Figure 4 illustrates this case. At zero earned income, the unemployed single  person  income unit received a maximum benefit of $160.10 pw in 1996. The difference between the Newstart Allowance recipient and the Sole Parent Pension  recipient at zero earned income reflects the greater support provided to the sole parent  income unit and to dependent children, under  the Australian  income support system. Earned income up to the disregard threshold point does not reduce entitlement. Beyond the disregard threshold point ($30  pw), the Newstart Allowance payment is reduced by 50 cents  in the dollar up to $70 pw and by 70 cents  in the dollar thereafter. The increase in the withdrawal rate from 50 per cent  to 70 per cent  ensures a flatter  Newstart Allowance disposable income line than for the sole parent  case. Notice also the effect of the tax system  in the case of the unemployed single  person  income unit. A net tax amount  is paid by the single Newstart Allowance recipient at $15 pw of earned income. This level  of earned income is well below the disregard threshold point. The METR at $15 pw moves from 0 to 20 per cent. The single  Newstart Allowance recipient pays tax at an earlier point than the Sole Parent Pension recipient for the same earnings profile  because the rebates applying to the Sole Parent Pension have a much  stronger effect in reducing effective tax paid at low income levels  than for the Newstart Allowance recipient. Moreover, the unemployed single  person  income unit is unaffected by the Family Tax Assistance package which raises  the threshold level  for payment of tax.

Entitlement to Newstart Allowance is lost at an earned income figure  of $271  pw; well  below the Sole Parent Pension  entitlement cut-off point.

The unemployed couple with dependent children

Figure 5 presents the case of the unemployed couple with  two dependent children. We assume that the Newstart Allowance recipient has a partner who  is in receipt of Parenting Allowance. For the purposes of this simulation we assume  that the partner is not earning income. Family Payment  is also paid to the income unit. At $30 pw, the Newstart Allowance recipient faces a withdrawal rate of 50 cents  in the dollar. Under the rules  applying to Parenting Allowance, the partner of the Newstart Allowance recipient faces no withdrawal of Parenting Allowance while the Newstart Allowance recipient retains  an entitlement to income support. At an earnings level  of $70 pw, the withdrawal rate for the Newstart Allowance recipient rises to 70 cents  in the dollar. At $58 pw, the unemployed couple income unit begins  to pay a net tax amount. The combined effect of a 20 per cent  net tax rate and a 70 per cent  withdrawal rate pushes  the METR to 90 per cent  beyond  the $70 per week point. At $248  pw, Newstart Allowance payment is exhausted. From this point on, Parenting Allowance is reduced at the withdrawal rate of 70 per cent. Hence, the METR remains at high levels. Finally at $408  pw, entitlement to the additional rate Parenting Allowance is lost. What remains is the basic  rate of Parenting Allowance, Family Payment  and, of course, the income unit’s  post-tax  earnings.

The earned income supplement and poverty alleviation

In the introduction to this report, we suggested that the provision of the earned income supplement supports the poverty alleviation objective of social  policy. We now  illustrate this connection. Figure 6 presents estimates of the ratio of disposable income to the poverty line. This ratio is evaluated at different  levels  of earned income. What we are interested in is the extent to which an increase in labour  earnings moves the income unit above their  relevant poverty line. The poverty line specified in figure  6 is the before-housing costs Henderson Poverty Line (HPL). Built into the HPL are the expenditure needs  of different  income unit types. The HPL for the single  unemployed income unit is lower than that for the couple with  two dependent children unemployed income unit and the sole parent  (two  dependent children) income unit because their  expenditure needs are lower.

At zero labour  earnings, all income units find themselves below the poverty line.17 The effect of the earned income supplement is to ensure that all income units move above the HPL with relatively low levels  of earned income (around $75 pw). The additional support provided to sole parents under  the tax-transfer system  is evident in the greater rise in the sole parent income unit’s  ratio of disposable income to the poverty line relative to the corresponding ratio for the unemployed income units.

Figure 6:  The ratio  of disposable income to the before-housing Henderson Poverty Line, end 1996

Figure 6:  The ratio  of disposable income to the before-housing Henderson Poverty Line, end 1996

Source: Authors’ simulation model results.

2.5 Summary

This chapter has pointed to the growing importance of part-time  and casual  work  in the Australian  labour  market. One supply-side factor influencing the growth in part-time employment is the structure of income tests. In easing  income test limits, more income support recipients are encouraged to take up short-hour  employment options. For some, a short-hours job is the optimal  outcome. For others, the short-hours  job is a means  to an end as to gain
full-time employment.

Our analysis of the affect of earned income on the contour of the METR schedules for each  of the case studies  examined, illustrates a number  of important principles.

First, METRs are very high at even  low levels  of private income for income support recipients. This is particularly the case for the unemployed recipient cases  analysed. In other  words, the ‘marginal cost’ of working sets in at very low levels  of earned income.

Second, different  income units, within the same income support payment category, may face different  incentive structures in regard  to the marginal dollar of earned income even  when  they face the same income test. The differences between the METR profiles  for the two unemployed income unit simulation cases  illustrate this principle. This difference eventuated because both entitlement amounts  and tax treatments differed  between the two cases.

Third, the allowance income test is considerably more severe  in its impact  than the pension income test. Receipt of the Sole Parent Pension  is maintained for a considerably longer  range of earned income than is the case of the two unemployed income unit case studies. This, of course, means  that we would expect a much  longer  tail in the distribution of earned income for sole parent  income units than unemployed income support recipients. In comparing the Sole Parent Pension  profile  with  the Newstart Allowance profile, it is evident that, behavioural effects aside, the more liberal  Sole Parent Pension  income test results  in a greater retention of part-time employed sole parents than would  be the case if the Newstart Allowance income test applied. In other  words, if the allowance income test applied to sole parent  pensioners, large  numbers of Sole Parent Pension  recipients in receipt of part-time  earnings would  lose entitlement. Correspondingly, the incidence of earned income receipt among  Sole Parent Pension  recipients would  fall, all other  things being  equal.

Fourth, in the case of dual-income support payment income units, the loss of entitlement to an income support payment for one recipient does not result  in the reduction of high METRs for the income unit generally. In the case of the unemployed couple with  dependent children case study, for example, high METRs continued to apply  following the loss of entitlement to Newstart Allowance. This is because the partner in receipt of Parenting Allowance began to lose entitlement at 70 cents  in the dollar of earned income (of the former Newstart Allowance recipient).

[ Return to Top   Return to Section ]

3 The data

3.1 The longitudinal data set (LDS) 1% Sample

3.2 Income support payment categories

3.1 The longitudinal data set (LDS) 1% Sample

The LDS 1% Sample  contains fortnightly records of a sample  of income support recipients and their  partners. The data used in this paper  refer to the period  July 1995  to June 1999. The data are drawn  from Centrelink’s administrative payment records system  and cover  the whole of the Australian  population. Personal  details  include: the date of birth of the person, their country of birth, Aboriginal  and Torres Strait Islander  status, housing tenure status, postcode, gender, education level  (for certain payment categories) and the number  of children they have in eligible Family Allowance categories. All variables are specified at the fortnight extract date point.

The LDS 1% Sample  contains details  on the private income of the person. Private  income is split between ‘earned income’ or the income derived from labour  market  activities (wages and salaries) and ‘unearned income’ (all other  income). The income support payment type  of the individual is listed  at the time of receipt of payment. Income  support recipients who  move between payment categories will  be recorded in separate income support payment categories for different  extract dates. The rate of payment is included in the LDS 1% Sample  together with maximum entitlement and add-ons such as rent assistance. The activity status of the recipient is provided in the case of Newstart Allowance and Youth Training/Youth  Allowance payment recipients. Education  details  are available for those  receiving income support in the Youth Allowance category.

One drawback of the LDS 1% Sample  data is that it only includes fortnightly information on individuals if those  individuals remain  on the Centrelink administrative system. Once off the system  we do not observe them. This means  that we do not know  the post-entitlement status of the individuals concerned. In particular, we do not know  whether an exit  from income support is an exit  to employment (or, more accurately, to an employment position with  sufficiently high earnings to preclude an income support payment) or to some non-employment state.

3.2 Income support payment categories

In this study, we focus on four unemployment-based and two family-based categories of income support recipients. Table 2 provides details  of the specific payment types  encompassed in these programs and the income tests that apply  to each category.

The first category of unemployed income support recipients brings  together all recipients in the LDS 1% Sample  who  are unemployed and under  the age of 21. The category is referred to as the Unemployed (<21) throughout the report. There are two sub-components of this group. The first includes recipients receiving support under  the former Youth Training  Allowance, the former  Job Search  Allowance, and Newstart Allowance. The second  comprises all unemployed persons under  the age of 21 receiving income support payments under  the Youth Allowance program.

Table 2: Income support payment categories, payment eligibility, and treatment of labour earnings
Category LDS 1% Sample payment Types Eligibility Treatment of labour earnings
Unemployed
(<21)
Youth Training Allowance (YTA) under 21. Job Search Allowance (JSA) under 21.
Newstart Allowance (NSA) under 21.
Unemployed registered with the CES in early . years of LDS 1% Sample.
Approved education, training and job search activities.
Waiting periods apply
.Mutual obligations requirements apply to recipients who have been unemployed for 6 months (aged 18–24) or 12 months (aged 24–34 years).
Disregard ($60 per fortnight)
.Benefit Reduction Brackets
—50 cents per dollar between $60 and $140 per fortnight
—70 cents per dollar over $140
Earnings Credit Scheme (abolished in March 1997). The income test is applied firstly to the income of the recipient.
Once the cut-off point is reached, earned income affects the receipt of an income support payment by the partner.
Unemployed
(21+)
Job Search Allowance (JSA) 21 and over.
Newstart Allowance (NSA) 21 and over.
Unemployed registered with the Commonwealth Employment Service (CES) in early years of LDS 1% Sample Activity agreements and compliance with activity test.
Waiting periods.
Mutual obligations requirements apply to recipients
who have been unemployed for 6 months (aged
18–24) or 12 months (aged 24–34 years)
As for Unemployed (<21).
Unemployed
(60+)
Mature Age Allowance (MAA)
Newstart Mature Age (NMA) .
Aged 60 years or over but less than pension age. No recent workforce experience
Received income support payment prior to
Mature Age Allowance claim. Not activity tested.
As for Unemployed (<21).Grants of MAA provided prior to 1 July 1996 are assessed under the pensions income test.
Partner Allowance Partner Allowance (PA and PTA)
Mature Age Partner Allowance (MPA)
Born on or before 1 July 1955. As for Unemployed (<21). Partner on a pension or allowance.
No recent workforce experience.
No dependent children under 16 years.
Not activity tested.
Parenting
Single
Parenting Payment Single (PPS)
Single Sole Parent Pension (SPP)
Not a member of a couple or living away from Legal custody of a dependent child. The disregard is adjusted each year. The disregard was partner. $94 per fortnight in 1 July 1995. In July 1999, the disregard was $102 per fortnight. Benefit Reduction Rate—50 cents per dollar.Earnings Credit Scheme (Abolished in March 1997)
Parenting
Partnered
Parenting Allowance Newstart Partner (PGN)
Parenting Allowance Low (PGL)
Parenting Allowance Youth
Allowance Partner (PGY)
Member of a couple. Have a dependent child. Income test dependent on both individual and combined income. Partner income affects the additional rate of Parenting Payment when partner income results in exhaustion of partner allowance (or equivalent levels of income if partner is not in receipt of an allowance).

From 1 July 1998, all eligible unemployed persons under  the age of 21 received payment under the broad Youth Allowance payment umbrella. The Youth Allowance payment also covers  young people who  are studying or who  are sick. A comparison of the former Youth Training  Allowance recipient series  and the Youth Allowance recipients identified in the LDS 1% Sample  as unemployed, suggests that the two series  are not perfectly compatible but that the degree of incompatibility is not large.

The second  income support payment category identified relates to unemployed income support recipients aged  21 and over. This category is identified throughout the study  as the Unemployed (21+). Under current arrangements, eligible unemployed persons 21 and over receive payment under  Newstart Allowance. Prior to 20 September 1996, Newstart Allowance was provided to long-term  unemployed people over the age of 18, while the Job Search  Allowance payment provided income support to unemployed people aged  18 or over,  unemployed for periods of less than a year. From 20 September 1996, Job Search  Allowance and Newstart Allowance were combined into a single  Newstart Allowance payment. To ensure consistency between current and historical payment arrangements, our analysis applies a 21 and over age restriction to both Newstart Allowance and Job Search  Allowance recipients.

The third category comprises older  unemployed income support recipients aged  60 and over, but below pension age. This category is identified throughout the study  as the Unemployed (60+). Unemployed recipients aged  60 and over receive the Mature Age Allowance payment. New entrants to Mature Age Allowance from 1 July 1996  have, however, been  assessed under different  income testing  arrangements to those  receiving payment prior to 1 July 1996. Pre-July 1996  Mature Age Allowance recipients were assessed under  the more generous pensions’ income and assets  tests rather  than the allowance income and assets  test that apply  post 1 July 1996.

The final unemployed income support payment recipient category identified is the Partner Allowance category. This category provides income support payments under  the Partner Allowance payment program to those  persons born on or before  1 July 1955  who  are partners of recipients and who  have no recent workforce experience.

Two Parenting Payment  categories are distinguished. The first category, Parenting Single, covers sole parent  income support recipients receiving Parenting Payment  Single payments or, prior to 20 March 1998, the Sole Parent Pension. There are minor differences between the two sole parent  income support payments. For example, the Sole Parent Pension  was assessed under  the pension assets  test while the Parenting Payment  Single payment is assessed under  the ‘sudden death’ allowance assets  test.

The second  Parenting Payment  category, Parenting Partnered, covers  low-income parents caring for children. The Parenting Payment  for those  partnered is divided  into a basic  Parenting Payment  component, which is assessed under  the claimant’s personal income position alone, and an additional Parenting Payment  component, which is assessed against  both the income of the recipient and the income of their  partner. Only income support recipients in receipt of more than the basic  rate of Parenting Payment  are covered in the Parenting Partnered category in our analysis. Those persons receiving only the basic  rate of Parenting Payment  are classified as non-income support recipients in the LDS 1% Sample  and are not included in the results which follow.

Strict unemployment and activity testing  eligibility requirements apply  with  respect to the Youth Training  Allowance, the Youth Allowance and Newstart Allowance while the Mature Age Allowance and Partner Allowance adopt  the more liberal ‘absence of recent workforce experience’ eligibility requirement. Maximum  rates of payment depend on age, independence status, and partner status.

3.3 Trends in selected payment categories

Trends in the number  of allowees and pensioners in the LDS 1% Sample  for the selected income support payment categories over the period  June 1995  to June 1999  are displayed in figures 7 to 10. The Unemployed (21+) category comprises the largest  group  (7,000–8,000) among the selected classes  of income support recipients in the LDS 1% Sample  followed by the Parenting Single category (around 3,500) and the Parenting Partnered group  (around 2,400). The remaining payment types  Unemployed (<21), Unemployed (60+) and Partner Allowance contain  relatively small numbers in the LDS 1% Sample. Inferences drawn  from the LDS 1% Sample  relating to these  particular payment types  must be regarded as tentative.

Figure 7: Unemployed  income support  payment  recipients, men, LDS 1% Sample

Figure 7: Unemployed  income support  payment  recipients, men, LDS 1% Sample

Source: LDS 1% Sample

Figure 8: Unemployed income support payment recipients, women, LDS 1% Sample

Figure 8: Unemployed income support payment recipients, women, LDS 1% Sample

Source: LDS 1% Sample

Figure 9: Parenting income support payment recipients, men, LDS 1% Sample

Figure 9: Parenting income support payment recipients, men, LDS 1% Sample

Source: LDS 1% Sample

Figure 10: Parenting income support payment recipients, women, LDS 1% Sample

Figure 10: Parenting income support payment recipients, women, LDS 1% Sample

Source: LDS 1% Sample

As is evident from the trends  in the number  of income support recipients in different  payment categories, significant gender differences exist  in the composition of the various  payment types. Women  make up by far the largest  share  of the Parenting Single, Parenting Partnered and the Partner Allowance groups. Men, on the other  hand, comprise a much  larger  share  of the remaining payment classification types. Very few women are in the Unemployed (60+); the Age Pension  applying at an earlier age for women over the timeframe covered in the LDS 1% Sample. Likewise, very few men are located in the Partner Allowance category.

The number  of Youth Training  Allowance Unemployed (<21) income support recipients drops off markedly with  the introduction of the Youth Allowance in 1998. The Youth Allowance Unemployed (<21) series  start from this point.

[ Return to Top   Return to Section ]

4 Employment rate trends and determinants

4.1 Earned income data in the LDS 1% Sample

4.2 Trends in the employment rate

4.5 Summary

In this chapter, we investigate the ways  in which income support recipients engage in the labour  market. Section  4.1 examines trends  in employment rates (the  proportion of recipients in paid employment) among  the selected income support programs over the 1995  to 1999 period. In section 4.2, we estimate a statistical model  of labour  earnings receipt among recipients where we include both socio-demographic determinants and income support arrangement determinants. Our aim is to understand what  factors  influence the incidence of paid employment. The model  is estimated at both the beginning of the sample  period  of the Longitudinal Data Set (LDS) 1% Sample  (1995) and the end point (1999). The chapter closes with  an analysis of the distribution of earnings among  recipients in the unemployed and families  with  children income support programs.

4.1 Earned income data in the LDS 1% Sample

The LDS 1% Sample  records the level  of earned income reported by an income support payment recipient on a fortnightly basis.18  Inspection of the earned income variable suggests that the data appear in good order  for the 1995  to 1999  period. There are, however, unique problem fortnights in terms  of the earned income data for the Unemployed (<21) and Unemployed (21+) recipient groups. These extract dates  are the first fortnights in January 1996, January  1997  and for June 1996. The level  of earned income can be expected to decrease for the first fortnights in January  given  the post-Christmas holiday effect on the receipt of earnings from casual  employment. However, as figures  11 and 12 show, the decline in the proportion of Unemployed (<21) and Unemployed (21+) recipients receiving earned income for these  particular fortnight appears exceptional. The case of the June 1996  estimates, for the Unemployed (<21) recipients, suggests a unique administrative error. Notice also the sharp break in the Parenting Partnered category series  in the first month of July 1996  when  the rate of earned income receipt increases from 6 per cent  to around  11 per cent.

Figure 11: Employment  rates, unemployed income support  payment  recipients, women, LDS 1% Sample

Figure 11: Employment  rates, unemployed income support  payment  recipients, women, LDS 1% Sample

Source: LDS 1% Sample

Figure 12: Employment rates, unemployed income support payment recipients, men, LDS 1% Sample

Figure 12: Employment rates, unemployed income support payment recipients, men, LDS 1% Sample

Source: LDS 1% Sample

Figure 13: Employment rates, parenting income support payment recipients, women,LDS 1% Sample

Figure 13: Employment rates, parenting income support payment recipients, women,LDS 1% Sample

Source: LDS 1% Sample

Figure 14: Employment rates, parenting income support payment recipients, men,LDS 1% Sample

Figure 14: Employment rates, parenting income support payment recipients, men,LDS 1% Sample

Source: LDS 1% Sample

Before analysing trends  in the incidence of earned income receipt among  LDS 1% Sample recipients, one further  feature  of the earned income data should  be identified. Notionally, all persons in the LDS 1% Sample  are directly either  in receipt of an income support payment or the partner of an income support payment recipient. However, it is possible to identify  a class of recipients whose benefit  level  is zero but who  are retained in the Centrelink register and, therefore, in the LDS 1% Sample  for a short period. Retention of such recipients appears to be more common  in the latter  years  of the LDS 1% Sample  than the earlier years. One reason  for this retention practice, obvious  from the data, is that a high earnings employment position may only be transitory. The number  of recipients in this category is relatively low among  Parenting Single  recipients but somewhat higher  in the Unemployed (<21) and Unemployed (21+) categories.

4.2 Trends in the employment rate

Trends in the employment rate  (the  proportion of income support recipients receiving income from labour  market  activities) are represented in figures  11 to 14. As noted  in Chapter 2, part-time  employment rates are considerably higher  for women than for men in the Australian labour  market. It is, therefore, not surprising to see higher  employment rates for women than men across  the various  income support payment categories.

Parenting Payment income support recipients

As illustrated in figure  13, Parenting Single income support recipients exhibit the highest employment rates among  the various  income support payment categories. In 1995, the proportion of female  Parenting Single recipients receiving earned income was around  26 per cent. By the end of the sample  period  in June 1999, the employment rate had increased to around 29 per cent. The proportion of male Parenting Single  recipients engaged in paid labour market  activities is much  lower than the corresponding figure  for women but has also grown significantly over the same time period; from around  13 per cent  in 1995  to 20 per cent  in 1999 (see  figure  14). In terms  of the Parenting Partnered recipient group, the proportion of female recipients in paid employment rises sharply to 11 per cent  in July 1996  but from this point, the employment rate falls to a little  under  10 per cent.

Unemployed income support recipients

For female  Unemployed (21+) income support recipients, the employment rate dropped marginally from around  24 per cent  in 1995  to around  22 per cent  at the end of the period. Among male Unemployed (21+) recipients, around  17 per cent  were in receipt of earned income in 1995. By June 1999, the employment rate had fallen to around  14 per cent. In the Unemployed (<21) category, the decline in the proportion of recipients in paid employment was even  more marked. For example, around  22 per cent  of female  Unemployed (<21) recipients were in paid employment in 1995. By the middle  of 1999 the employment rate had dropped to around  14 per cent.

The receipt of earned income is very low among  those  in the Unemployed (60+) category. The proportion of income support recipients in this category who  receive earned income lies around  5 per cent; a figure  that approximates the paid employment rate for aged pensioners. The incidence of earned income among  Partner Allowance recipients is somewhat higher  than that for the Unemployed (60+) group. This may reflect  both a gender effect
and younger age cohort. In both cases  there  has been  a declining trend  in earned income receipt over time.

Explaining trends in the employment rate

As emphasised in Chapter 2, the proportion of persons employed on a part-time  basis in the Australian  labour  market  rose throughout the period  of the LDS 1% Sample. The trend  of a growing number  of female  Parenting Single income support recipients receiving earned income is roughly in line with  this growth trend  and consistent with  higher  levels  of voluntary part-time  employment among  sole parents. As is clear, however, from the above discussion, the same cannot  be said of the other  categories. If anything, the estimates suggest that there  has been, for the remaining categories, a slight  decline in the proportion of recipients in paid employment while on income support over the period  of the LDS 1% Sample.

The slight  fall in the proportion recipients in the four unemployed income support payment categories and the Parenting Partnered category who  take up paid employment while on income support may reflect  a number  of possible effects. First, the state of the labour  market will  obviously affect these  trends. While  employment growth occurred during  the period  1995 to 1997, the unemployment rate actually slightly rose in this period. Unemployed recipients may have experienced greater difficulties in obtaining jobs (including part-time  positions) over this period  than Parenting Single recipients. The unemployment position improved significantly in 1998. The improving state of the labour  market  may have resulted in a greater take-up of longer- hour jobs among  the unemployed rather  than shorter-hour jobs as may have been  the case for Parenting Single recipients. Unemployed income support recipients who  may otherwise have accepted short-hours  employment positions may instead  have moved  off income support altogether and into positions with  longer  hours and higher  pay. Second, the reduced availability of labour  market  programs providing subsidised employment for unemployed recipients as a result  of the winding back of Working Nation initiatives in 1997  and 1998  may have effected the availability of subsidized part-time  employment positions. Finally, mention should  be made of changes to the structure of income support arrangements over the LDS 1% Sample  period. During the period, the Earnings Credit  Scheme was abolished (in 1997), the Mature Age Allowance income test was aligned with  that for other  allowees, and the withdrawal brackets for non-pension income support payments remained unadjusted.

The Earnings Credit  Scheme enabled those  income support recipients receiving maximum rate of entitlement to build up an earnings credit  accrual which could  be used to offset earned income and so limit the effect of higher  labour  income on benefit  loss. The scheme provided an incentive, in addition  to the disregard and the less than 100 per cent  withdrawal of benefit, for income support recipients to engage in the labour  market. The removal  of the Earnings Credit Scheme in 1997, therefore, acted  to reduce the proportion of recipients in paid employment (all other  things  being  equal).19 However, the removal  of the Earnings Credit  Scheme could  be expected to only have a one-off negative impact  on the employment rate.

In terms  of income support structural influences on the employment rate, the non-indexation of withdrawal rates for non-pension payments deserves, perhaps, greater attention. The pensions’ income test provides for indexation of benefit  reduction brackets; the allowance income test does not. The non-indexation of benefit  reduction brackets has a similar  effect and acts in an analogous manner  to the better-known example of the non-indexation of tax brackets. The non-indexation of tax brackets results  in ‘bracket creep’ when  nominal  income rises. With a rise in nominal  income (but no rise in real income), persons who  formerly paid no tax will  now pay some tax, while some taxpayers who  formerly paid tax at one rate move to a higher  tax bracket and, consequently, higher  rate of tax. Likewise, allowees who  face non-indexed benefit withdrawal brackets may find that a rise in labour  earnings may force them out of the disregard region  and into the 50 per cent  withdrawal rate bracket. Similarly, income support recipients facing  a 50 per cent  withdrawal rate but on the margin  of the 70 per cent  withdrawal rate bracket may find that an increase in labour  earnings now  results  in 70 per cent  withdrawal of income support. Finally, some income support recipients will  find that a rise in income will result  in a loss of entitlement altogether, whereas, if withdrawal rate brackets were indexed, no loss would  occur. The conclusion that can be drawn  from this is that, over the 1995  to 1999 time period, the non-indexation of withdrawal rates for allowees, has acted  to reduce, all other things  being  equal, the employment rate of recipients in the categories covered by the allowance income test.

The indexation of withdrawal brackets for payments made  to sole parents provides one explanation for the higher  labour  income rate for the Parenting Single income support payment category evident from figures  13 and 14. To this should  be added  the combination of the more liberal  disregard and lower withdrawal rates for these  recipients together with  greater preferences for short-hours  employment. Abstracting from any behavioural effects, the more liberal  disregard and withdrawal rate structure for sole parents implies a low attrition rate for sole parents receiving labour  income than the other  income support payment types  considered in this report.

4.3 The determinants of labour earnings receipt

In the previous section, we found that the incidence of labour  earnings receipts was highest among  Parenting Single income support recipients. This result  applied throughout the sample period  of the LDS 1% Sample  for both women and men. We now  ask whether this result continues to hold after controlling for a range  of factors  that might  affect the incidence of labour  earnings among  recipients.

We model  the receipt of labour  earnings while on income support employing a logit model structure. The dependent variable takes  the value  of one if the income support recipient is employed and receiving labour  earnings (but still on an income support system); zero otherwise. Let X be the vector  of explanatory variables and p the probability that Y =1 conditional on X. Thus, under  the logistic distribution model:

Logit (p) = log (p/(1–p)) = B’X.

Separate regressions are estimated for women and men and for two time periods; namely,
July 1996  and June 1999. A set of socio-demographic variables is utilised together with  variables summarising the income support payment arrangements of the recipient.

Socio-demographic variables include:

  • Marital status (Married);
  • Number of dependent children and their  age structure (Number of dependent children, Aged 0–6, Aged 7–12, Aged 13–15, Aged over 15);
  • Country  of birth (Main English speaking country, Non-English speaking country; the default  is Australian  born);
  • Age (Age and Age squared);
  • Aboriginal  and Torres Strait Islander  status (Aboriginal and Torres Strait Islander status);
  • Housing tenure (Home purchaser, Other home owner, Private rent, Government rent, Other rent; the default  is outright home owner).

Income  support arrangement variables include:

  • Recipient payment type  (Unemployed (<21), Unemployed (60+), Partner Allowance, ParentingSingle, ParentingPartnered; the default  is Unemployed (21+));
  • Recipient payment levels  (Maximum income support entitlement, Rent assistance, Remote area allowance paid);
  • Partner  payment type  (Partner receives unemployment-related income support payments, Non-payment partner);
  • Unearned income;
  • Job search activity;
  • Duration on income support payments;
  • Duration of unemployment

Our results  for the years  1996  and 1999  are set out in tables  3 and 4. Having controlled for socio-demographic effects  (for example, age and gender), our results  reveal  a highly  significant positive relationship between the Parenting Single income support payment category and the incidence of labour  earnings. This association applies for both the 1996  and 1999  time points and for both women and men. The results  indicate that Parenting Single income support recipients have a higher  probability of being  employed while on income support than do other  recipients. This results  from the joint effect of sole parent  status and pension income test coverage.

In terms  of socio-demographic effects, male married income support recipients have a higher probability of earnings receipt in 1999  (but not in 1996). The higher  the number  of dependent children, the lower the probability of earnings receipt among  female  recipients. A positive (but diminishing) relationship exists  between the age of the recipient and labour earnings receipt.

The probability of Aboriginal  and Torres Strait Islander  pensioners and beneficiaries obtaining employment positions while on income support is considerably below that for the non- Indigenous population. Likewise, income support recipients who  are from a non-English speaking background also have a significantly lower probability of being  in paid employment. These results  are consistent with  the general finding  of disadvantage felt by these  two groups in the Australian  labour  market.

Several  other  results  are worthy of comment. First, a strong  negative relationship exists  between government rental  status and labour  earnings receipt. Government rental  status may act as a proxy  for long-term  disadvantage. The basis for this view  is that long waiting lists apply  to this particular housing sector. Moreover, entry  is based  on the application of a means  test. On this interpretation, long-term  disadvantage reflects itself in a lower probability of gaining paid employment positions. Another possible interpretation of the relationship between government rental  status and labour  earnings receipt is possible. Low rents and stability in lease arrangements characterise the government rental  sector  relative to the private rental  sector. Moreover, eligibility is based  on the continued meeting of means  test requirements. Together, these  factors  may reduce the incentive for some income support recipients to engage with  the labour  market. On the basis of these  arguments, we would expect recipients in the government rental  sector  to also have a lower conditional probability of exit  from the income support system. We return  to this question in Chapter 6 where we examine the question of exit outcomes for recipients. A second  interesting result  is that recipients whose Activity  Test status is designated as ‘job search  activity’ in the LDS 1% Sample  have a considerably higher  probability of receiving labour  income as compared with  other  recipients. incidence of labour  earnings. This association applies for both the 1996  and 1999  time points and for both women and men. The results  indicate that Parenting Single income support recipients have a higher  probability of being  employed while on income support than do other  recipients. This results  from the joint effect of sole parent  status and pension income test coverage.

In terms  of socio-demographic effects, male married income support recipients have a higher probability of earnings receipt in 1999  (but not in 1996). The higher  the number  of dependent children, the lower the probability of earnings receipt among  female  recipients. A positive (but diminishing) relationship exists  between the age of the recipient and labour earnings receipt.

The probability of Aboriginal  and Torres Strait Islander  pensioners and beneficiaries obtaining employment positions while on income support is considerably below that for the non- Indigenous population. Likewise, income support recipients who  are from a non-English speaking background also have a significantly lower probability of being  in paid employment. These results  are consistent with  the general finding  of disadvantage felt by these  two groups in the Australian  labour  market.

Several  other  results  are worthy of comment. First, a strong  negative relationship exists  between government rental  status and labour  earnings receipt. Government rental  status may act as a proxy  for long-term  disadvantage. The basis for this view  is that long waiting lists apply  to this particular housing sector. Moreover, entry  is based  on the application of a means  test. On this interpretation, long-term  disadvantage reflects itself in a lower probability of gaining paid employment positions. Another possible interpretation of the relationship between government rental  status and labour  earnings receipt is possible. Low rents and stability in lease arrangements characterise the government rental  sector  relative to the private rental  sector. Moreover, eligibility is based  on the continued meeting of means  test requirements. Together, these  factors  may reduce the incentive for some income support recipients to engage with  the labour  market. On the basis of these  arguments, we would expect recipients in the government rental  sector  to also have a lower conditional probability of exit  from the income support system. We return  to this question in Chapter 6 where we examine the question of exit outcomes for recipients. A second  interesting result  is that recipients whose Activity  Test status is designated as ‘job search  activity’ in the LDS 1% Sample  have a considerably higher  probability of receiving labour  income as compared with  other  recipients.

Table 3: Logit model, receipt of labour earnings, LDS 1% Sample, July 1996
  Males Females
  Parameter Wald Pr> Odds Parameter Wald Pr> Odds
Variable Estimate Chi-Square Chi-Square Ratio Estimate Chi-Square Chi-Square Ratio
Constant –3.434 39.09 0.00   –6.712 169.54 0.00  
Marital status/dependent children
Married 0.180 1.55 0.21 1.20 –0.082 0.22 0.64 0.92
Aged 0–6 0.285 1.85 0.17 1.33 0.094 0.35 0.55 1.10
Aged 7–12 0.203 1.08 0.30 1.23 0.323 5.38 0.02 1.38
Aged 13–15 0.299 3.50 0.06 1.35 0.445 16.40 0.00 1.56
Aged over 15 –0.290 0.07 0.79 0.75 0.319 0.29 0.59 1.38
Number of dependent children –0.016 0.05 0.82 0.98 –0.261 29.57 0.00 0.77
Age                
Age 0.033 1.70 0.19 1.03 0.234 91.47 0.00 1.26
Age Squared 0.000 1.29 0.26 1.00 –0.003 80.91 0.00 1.00
Country of birth                
Main English speaking  country –0.018 0.02 0.88 0.98 –0.077 0.53 0.47 0.93
Non English speaking  country –0.381 14.03 0.00 0.68 –0.802 72.78 0.00 0.45
Aboriginal and Torres  Strait Islander status
Aboriginal and Torres Strait Islander –0.899 8.27 0.00 0.41 –1.092 25.85 0.00 0.34
Housing tenure
Home purchaser –0.023 0.01 0.93 0.98 –0.115 0.44 0.51 0.89
Other home owner 0.119 0.16 0.69 1.13 0.179 1.36 0.24 1.20
Private rent 0.078 0.54 0.46 1.08 –0.018 0.02 0.88 0.98
Government rent –0.550 9.19 0.00 0.58 –0.660 38.28 0.00 0.52
Other rent –0.234 4.19 0.04 0.79 –0.271 6.74 0.01 0.76
Payment  type
Unemployed (< 21 JSA/YTA) –0.172 0.97 0.33 0.84 0.868 32.80 0.00 2.38
Unemployed (60+) –0.187 0.29 0.59 0.83 na na na na
PartnerAllowance 1.113 5.84 0.02 3.04 0.249 0.62 0.43 1.28
ParentingSingle 1.174 14.85 0.00 3.23 0.839 9.94 0.00 2.32
ParentingPartnered 0.545 2.46 0.12 1.72 1.083 12.93 0.00 2.96
Other variables
Maximum income  support entitlement 0.000 0.56 0.46 1.00 0.004 170.58 0.00 1.00
Rent assistance –0.372 11.46 0.00 0.69 –0.426 17.67 0.00 0.65
Remote area allowance –0.565 2.42 0.12 0.57 0.348 2.55 0.11 1.42
Partner receives unemployment-related IS payments 0.368 3.78 0.05 1.45 –0.220 1.75 0.19 0.80
Non-payment partner 0.447 3.19 0.07 1.56 –0.054 0.09 0.77 0.95
Unearned income –0.003 5.56 0.02 1.00 –0.005 22.28 0.00 1.00
Job search activity 1.053 25.53 0.00 2.87 0.614 8.46 0.00 1.85
Duration on income  support payments –0.062 1.27 0.26 0.94 –0.111 5.18 0.02 0.90
Duration of unemployment 0.085 2.25 0.13 1.09 0.072 1.79 0.18 1.08
N
Model fit and testing global null hypothesis
6573
Intercept
Intercept and Chi-square for covariates 8566
Intercept
Intercept andChi-square for covariates
Akaike Information Criterion only
5658.7
covariates
5474.5
only
8815.4
covariates
7876.2
Schwarz Criterion 5665.5 5685.0 8822.5 8087.8
– 2 LOG L Score test 5656.7 5412.5 244.205 with 30 DF (p=0.0001)
226.493 with 30 DF (p=0.0001)
8813.4 7816.2997.238 with 29 DF (p=0.0001)
883.267 with 29 DF (p=0.0001)

Source: LDS 1% Sample.


Table 4: Logit model, receipt of labour earnings, LDS 1% Sample, June 1999
  Males Females
  Parameter Wald Pr> Odds Parameter Wald Pr> Odds
Variable Estimate Chi-Square Chi-Square Ratio Estimate Chi-Square Chi-Square Ratio
Constant –3.951 52.82 0.00   –3.995 75.44 0.00  
Marital status/dependent children
Married 0.496 11.78 0.00 1.64 –0.022 0.02 0.89 0.98
Aged 0–6 0.244 1.25 0.26 1.28 –0.224 2.09 0.15 0.80
Aged 7–12 0.149 0.54 0.46 1.16 0.175 1.71 0.19 1.19
Aged 13–15 0.374 4.65 0.03 1.45 0.458 18.13 0.00 1.58
Aged over 15 –0.012 0.00 0.96 0.99 0.319 5.10 0.02 1.38
Number of dependent children –0.185 5.65 0.02 0.83 –0.252 28.14 0.00 0.78
Age
Age 0.106 14.21 0.00 1.11 0.162 46.28 0.00 1.18
Age Squared –0.001 14.46 0.00 1.00 –0.002 44.85 0.00 1.00
Country of birth
Main English speaking  country –0.068 0.23 0.63 0.93 –0.061 0.34 0.56 0.94
Non English speaking  country –0.343 10.24 0.00 0.71 –0.799 78.42 0.00 0.45
Aboriginal and Torres  Strait Islander status
Aboriginal and Torres Strait Islander –1.410 15.93 0.00 0.24 –0.999 23.65 0.00 0.37
Housing tenure
Home purchaser 0.096 0.21 0.65 1.10 0.339 6.86 0.01 1.40
Other home owner 0.509 2.22 0.14 1.66 0.219 1.42 0.23 1.25
Private rent 0.008 0.01 0.94 1.01 0.088 0.71 0.40 1.09
Government rent –0.446 5.96 0.01 0.64 –0.755 51.67 0.00 0.47
Other rent –0.665 20.18 0.00 0.51 –0.400 8.88 0.00 0.67
Payment  type
Unemployed  (< 21 JSA/YTA) –0.758 1.56 0.21 0.47 –0.425 0.88 0.35 0.65
Unemployed  (< 21 YA) 0.426 6.20 0.01 1.53 0.090 0.38 0.54 1.10
Unemployed  (60+) –0.430 1.84 0.18 0.65 na na na na
Partner Allowance –0.681 1.15 0.28 0.51 –0.205 0.82 0.37 0.82
Parenting Single 0.858 12.53 0.00 2.36 0.938 22.01 0.00 2.56
Parenting Partnered –0.217 0.59 0.44 0.81 –0.140 0.30 0.59 0.87
Other variables
Maximum income  support entitlement 0.031 0.41 0.52 1.03 0.084 14.51 0.00 1.09
Rent assistance –0.280 6.20 0.01 0.76 –0.322 10.83 0.00 0.72
Remote area allowance –0.757 0.68 0.41 0.47 –2.430 13.12 0.00 0.09
Partner receives unemployment-related IS payments –0.155 0.56 0.45 0.86 –0.326 5.33 0.02 0.72
Non-payment partner 0.203 0.60 0.44 1.23 –0.152 0.90 0.34 0.86
Unearned income –0.003 3.09 0.08 1.00 –0.005 21.02 0.00 1.00
Job search activity 0.472 26.98 0.00 1.60 0.538 26.27 0.00 1.71
Duration on income  support payments –0.017 0.19 0.67 0.98 –0.057 1.85 0.17 0.94
Duration of unemployment 0.023 0.35 0.55 1.02 0.058 1.87 0.17 1.06
N 6484       9029      
Model fit and testing global null hypothesis                                               Intercept         Intercept andChi-square for covariatesIntercept         Intercept andChi-square for covariates
  only covariates only               covariates  
Akaike Information Criterion 5137.0 4944.3 9183.8                  8317.2  
Schwarz Criterion 5143.7 5161.2 9190.9                  8537.5  
– 2 LOG L 5135.0 4880.3254.641 with 31 DF (p=0.0001)9181.8 8255.2926.658 with 30 DF (p=0.0001)  
Score test   230.222 with 31 DF (p=0.0001) 844.063 with 30 DF (p=0.0001)  

Source: LDS 1% Sample.

4.4 The distribution of labour earnings

Having examined the receipt of labour  earnings among  pensioners and beneficiaries we now examine the distribution of earnings among  those  in receipt of earnings. The distribution of earnings for male and female  income support recipients in the six identified categories is contained in figures  15 to 20. The figures  illustrate the distribution of earnings both at the beginning of the sample  period  in 1995  and at the end of the sample  period  1999.20 Table 5 below contains estimates of median  earnings for women and men across  the various  payment categories. Our discussion is presented in brief form, as the analysis of earnings distributions, utilising the LDS 1% Sample, is the subject of a separate paper  by Creedy  & Scutella  (2000).

The pension income test allows  income support recipients to earn considerably more income per fortnight than the allowance test before  entitlement to benefit  is lost. Recall  from our simulation exercise for 1996  in Chapter 2 that entitlement to Newstart Allowance was lost at $540  per fortnight.21 In the case of Parenting Single recipients, entitlement to the Parenting Payment  for sole parents is not lost until the recipient earns  $838  per fortnight. We would expect this difference to show  up in estimates of the distribution of earnings between different recipient groups. Figures  15 to 20 confirm  this result. The earnings distribution profile  for Parenting Single  recipients is skewed well  to the right of the distribution of earnings for all other  income support payment groups.22

Table 5: Median earnings of women and men, LDS 1% Sample.
  July  1995 July  1996 July  1997 July  1998 June  1999
Unemployed (< 21)
Women 149.5 157.5 150.0 147.5 150.0
Men 107.0 124.5 140.0 161.0 171.5
Unemployed (21+)
Women 162.5 200.0 170.0 282.0 209.0
Men 166.0 180.0 160.0 251.0 165.0
PartnerAllowance
Women 150.0 181.5 181.5 207.0 193.5
Unemployed (60+)
Men 100.0 156.0 150.0 135.0 152.0
ParentingSingle
Women 319.0 340.0 349.0 339.0 357.0
Men 150.0 244.5 206.0 293.0 210.0
ParentingPartnered
Women 199.0 200.0 180.0 197.0 210.5
Men 95.0 222.0 198.0 117.5 190.0

Source: LDS 1% Sample.

Figure 15: The distribution of earnings, Unemployed  (< 21), LDS 1% Sample

Figure 15: The distribution of earnings, Unemployed  (< 21), LDS 1% Sample

Figure 16: The distribution of earnings, Unemployed (21+), LDS 1% Sample

Figure 16: The distribution of earnings, Unemployed (21+), LDS 1% Sample

Figure 17: The distribution of earnings, Unemployed (60+), LDS 1% Sample

Figure 17: The distribution of earnings, Unemployed (60+), LDS 1% Sample

Figure 18: The distribution of earnings, Partner Allowance, LDS 1% Sample

Figure 18: The distribution of earnings, Partner Allowance, LDS 1% Sample

Figure 19: The distribution of earnings, Parenting Single, LDS 1% Sample

Figure 19: The distribution of earnings, Parenting Single, LDS 1% Sample

Compare, for example, the estimates contained in figure  19 for Parenting Single  recipients with those  for Unemployed (21+) recipients in figure  16. In the case of the Parenting Single recipients, around  39 per cent  of recipients were earning more than $400  per fortnight in June 1995. That proportion had risen  to 44 per cent  in June 1999. At the same time, median  earnings for female  (male) Parenting Single  recipients were $319  ($150) per fortnight in July 1995  and $359 ($210) per fortnight in June 1999. Among Unemployed (21+) recipients, on the other hand, only 11 per cent  of recipients were earning more than $400  per fortnight in June 1995 while 15 per cent  were in this earnings bracket in June 1999. Median earnings for female (male) Unemployed (21+) recipients were also well  below the estimates for Parenting Single recipients at $163  ($166) per fortnight in July 1995  and $209  ($165) per fortnight in June 1999.

Figure 20: The distribution of earnings, Parenting  Partnered, LDS 1% Sample

Figure 20: The distribution of earnings, Parenting  Partnered, LDS 1% Sample

Earnings for the Unemployed (<21) recipients are, not surprisingly, the lowest of all the categories. Median earnings for female  (male) Unemployed (<21) recipients were $150 ($107) per fortnight in July 1995  and $150 ($172) per fortnight in June 1999. The lower earnings of the Unemployed (<21) group  reflect  both an age-related influence and the effect of the additional parental income test on entitlement. Interestingly, earnings are also low at the opposite end of the age structure; namely, among  the Unemployed (60+) group. The median earnings for male Unemployed (60+) recipients were $100  per fortnight in July 1995  and $152  per fortnight in June 1999. The tenuous link with  the labour  market  among  this group represents a continuing theme  of our paper. Among female  Partner Allowance recipients, median  earnings grew  from $150  per fortnight in July 1995  to $194  per fortnight in June 1999.

A significant gender effect is evident in the earnings data: earnings for male income support recipients are generally, but not always, below those  for their  female  counterparts. This represents a reversal of what  we find in the labour  market  generally, where a significant gender wage  differential in favour of men is evident. In the case of Unemployed (21+) recipients, the median  earnings differential in favour of women is $45 per fortnight in June 1999  while for Parenting Single  recipients the differential in favour of women is $147  per fortnight. In the case of the Parenting Partnered recipients, the gender wage  differential (in favour of women) in June 1999  was $21. Only in the case of the Unemployed (<21) group, was the earnings differential in favour of men ($21  per fortnight).

One qualification must be noted  at this point. The LDS 1% Sample  does not include information on hours worked per week. And so our estimates of earnings differentials do not control  for hours of employment. This means  that we can say nothing of gender hourly wage  differentials from the LDS 1% Sample. Discussion  of the gender wage  differential is often made  in relation to hourly  wages rather  than fortnightly earnings (as is the case in the present analysis).

The final point of interest in terms  of earnings is in the question of how  income test arrangements affect incentives to engage in the labour  market. As explained in Chapter 2, income support recipients face a relatively high benefit  withdrawal rate (50 per cent)  when earnings pass the relevant disregard point. For recipients covered by the allowance income test, the withdrawal rate increases even  further  to 70 per cent  at higher  earning levels  ($140). Does the movement from a zero withdrawal rate to a 50 per cent  (and then again  to 70 per cent) withdrawal rate mean  that some pensioners and beneficiaries have a tendency to remain  at lower earnings levels  than they  may otherwise? In other  words, do we see a degree of bunching at points  of sharp  increase in withdrawal rates?23 The evidence presented in table  5 reveals  only a minor degree of bunching for the June 1999  earnings distributions at the $60 per fortnight disregard point in the case of Unemployed (<21) and Unemployed (21+) recipients but little evidence for withdrawal rate effects  elsewhere.

4.5 Summary

The employment rate in this study  is defined  as the proportion of income support recipients who  take up employment positions while on income support. Evidence  from the LDS 1% Sample  indicates that the employment rate is indeed highest among  Parenting Single recipients. The proportion of female  Parenting Single  recipients receiving earned income was around  26 per cent  in 1995  but had risen  to 29 per cent  in June 1999. As a point of comparison, the employment rate among  female  Unemployed (21+) recipients, was not only lower but had also fallen over the period  covered by the LDS 1% Sample; from 24 per cent  in
1995  to around  to 22 per cent  at the end of the period. A similar  decline in the employment rate, but from a lower initial  base, is evident among  male Unemployed (21+) recipients. Among Unemployed (<21) recipients, the decline in the proportion of recipients in paid employment was even  more pronounced over the 1995  to 1999  period. Various explanations for this trend were put forward  in the paper, including higher  exits  from the state of unemployment into
full-time employment towards the end of the sample  period, and the non-indexation
of withdrawal brackets for allowees. The take-up of employment positions among  older unemployed recipients has always been  very low and remains an obvious  concern for policy makers.

Modelling of the take-up of employment positions while on income support indicates that Parenting Single  recipients have a higher  probability of being  employed while on income support than do other  income support recipients having  controlled for other  determinants of earned income receipt. These determinants include the number  of dependent children in the case of female  recipients (negative effect), age (positive, but diminishing effect), Aboriginal  and Torres Strait Islander  status (negative), non-English speaking background (negative), government rental  status (negative), and job search  activity under  the Activity  Test (very  strongly positive).

[ Return to Top   Return to Section ]

5 Income support spells and employment episode histories

Each income support recipient in the Longitudinal Data Set (LDS) 1% Sample  is allocated a unique identifier number  (the  Customer Reference Number). This identifier stays with  the recipient over time. This means  that we can derive  an income support spell  and employment episode history  (or logbook) for each  recipient in the LDS 1% Sample  over the period  1995  to 1999. Throughout this paper, the term ‘income support spell’ denotes a continuous period  on income support (or, more precisely, a continuous period  on the Centrelink register). The term ‘employment episode’ is used in this paper  to refer to a continuous period  of paid employment while on income support. The LDS 1% Sample  follows  the history  of the recipient only when the recipient is on the Centrelink register. It does not provide  employment information when the recipient is not on the register. This means  that the term ‘employment episode’ does not refer to a period  of employment when  an individual has exited the income support system. By definition an employment episode can only be experienced when  the individual remains on the income support system.

The histories profile  of each  income support recipient allows  us to estimate the duration of each income support spell  and employment episode together with  the frequency of income support spells  and employment episodes over the 1995  to 1999  period. Section  5.1 provides summary details  of the income support spell  and employment episode histories across  all years of the LDS 1% Sample. Recipients may enter  the Centrelink register at any time in the course  of the LDS 1% Sample  and so our analysis of histories will  be made  up of very different  groups  of income support recipients. To control  for this, we refine  the analysis, in section 5.2, by focusing on a particular cohort  of income support recipients; namely, the 1995  group. The final section undertakes a transition analysis. This analysis checks a cohort’s income support payment and exit status at yearly intervals. Transitions may include the movement between one income support payment category to another  category and between the states  of ‘on income support’ and ‘off income support’.

5.1 Income support spells and employment episodes

Income support spells

An income support spell  is defined  for our purposes as a period  of continual observation in the LDS 1% Sample  in any one of the six identified income support payment categories. Our definition of an income support spell  is not based  on the receipt of an actual  payment of a benefit  in a particular fortnight. It is simply  based  on whether the income support recipient is maintained in the Centrelink administrative register under  a nominated income support payment type. As indicated earlier, it appears that recipients may be maintained in the Centrelink administrative register for a short period  even  if earnings reduce benefit  to zero. It should  also be noted  that the definition of an income support spell  allows  for movement between income support payment codes  for the six nominated groups  of recipients; namely, the Unemployed (<21), Unemployed (21+), Unemployed (60+), Partner Allowance, Parenting Single and Parenting Partnered categories. A spell  ends if there  is a transition from a nominated category code  to other  payment type  (for example, from the Mature Age Allowance to the Age Pension), a transition from income support payments to non-income support payments (for example, from the Newstart Allowance to the basic  rate of Parenting Payment), or an exit  from the income support system  altogether. For the present purpose, an exit  occurs  if we do not see the recipient in the LDS 1% Sample  for a given  fortnight when  we had seen  them in the data set in the previous fortnight.

The analysis below is based  on income support spells  for the entire  period  of the LDS 1% Sample. Importantly, the analysis is based  on both completed and uncompleted spells  of income support. Spells existing in the data in the first fortnight of the LDS 1% Sample  are also included. In more formal language, the analysis in this chapter includes both left-censored (or pre- existing) income support spells  and right-censored spells  (or spells  of income support that continue beyond  the end of the sample  period).

Our chief  point of focus is on the history  of employment episodes. Before examining this issue, however, we first present information on the profile  of income support spells  themselves. Figures  21 and 22 present estimates of the duration distribution of income support spells. We divide  spells  into five duration categories. The first category includes spells  of income support lasting  less than three  months. The second  category includes spells  lasting  between three  and six months. The third, fourth and fifth categories include spells  lasting  between six and twelve months, between twelve months  and a year, between one and two years, and over two years.

Figure 21: The distribution of income support  spells by payment  category, women, LDS 1% Sample

Figure 21: The distribution of income support  spells by payment  category, women, LDS 1% Sample

Source: LDS

Figure 22: Distribution of income support  spells by payment  category, men, LDS 1% Sample

Figure 22: Distribution of income support  spells by payment  category, men, LDS 1% Sample

Source: LDS 1% Sample

As is apparent from figures  21 and 22, the distribution of income support spells  for the two main unemployed categories (the  Unemployed (<21) and Unemployed (21+) groups) is quite different  to that which obtains  for other  income support categories. Short spells  are more prevalent for these  categories. Long income support spells  are less common. Income  support spells  of less than three  months  comprise around  40 per cent  of all income support spells for the Unemployed (<21) and Unemployed (21+) groups. In sharp  contrast, spells  of over 24 months  comprise less than 10 per cent  of all spells  for these  two groups  of income support categories.

A quite  different  profile  is evident in the case of the two remaining unemployed categories; namely, the Partner Allowance and Unemployed (60+) categories. Here, short spells  of income support comprise less than 20 per cent  of all spells. The same profile  is also evident in the case of Parenting Single income support payment recipients. For female  Parenting Single recipients, short income support spells  only account for 20 per cent  of all spells. The income support spell experience of Parenting Partnered recipients lies between that for the Unemployed (<21) and Unemployed (21+) recipients on the one hand and the profile  for Parenting Single recipients on the other.24

Employment episodes

We now  turn to the analysis of the employment episode experience of income support recipients. We remind  the reader that an ‘employment episode’ refers  to an episode of paid employment while on income support. Table 6 presents the distribution of employment episodes classified by the duration of income support spells. Table 7 provides estimates of the mean  duration of employment episodes and the proportion of time spent  in paid employment among  those  income support recipients who  gain employment (while on income support).

Table 6: The distribution of employment episodes by payment category, LDS 1% Sample.
  Unemployed (<21) Unemployed (21+) Unemployed (60+) Partner Allowance Parenting Single Parenting Partnered
Women Men Women Men Women Men Women Men Women Men Women Men
Numberof employment episodes income support spells of less than 3 months
None 64.5 76.4 62.6 72.7 * 76.9 83.9 * 58.4 73.8 84.1 84.1
One 32.0 20.9 32.7 24.4 * 15.4 15.3 * 40.9 26.2 14.9 14.9
Two or more 3.5 2.8 4.7 2.9 * 7.7 0.8 * 0.7 0.0 1.0 1.0
Total 100.0 100.0 100.0 100.0 * 100.0 100.0 * 100.0 100.0 100.0 100.0
Number  of employment episodes                                              Income support spells of 3 to less than  6 months
None 53.7 65.9 46.1 54.2 * 83.3 81.3 * 45.5 52.9 76.4 75.5
One 29.4 23.1 31.3 27.7 * 16.7 11.9 * 50.5 43.1 17.8 18.2
Two 13.7 8.1 17.8 14.0 * 0.0 5.7 * 3.5 3.9 3.9 3.6
Three 2.8 2.3 4.2 3.5 * 0.0 0.6 * 0.4 0.0 1.8 1.8
Four or more 0.4 0.5 0.6 0.6 * 0.0 0.6 * 0.0 0.0 1.9 0.9
Total 100.0 100.0 100.0 100.0 * 100.0 100.0 * 100.0 100.0 100.0 100.0
Number  of employment episodes                                            Income support spells of 6 to less than  12 months
None 48.0 60.0 40.9 47.9 * 85.0 77.4 * 46.4 57.4 75.3 71.0
One 22.1 20.6 22.8 21.0 * 5.0 12.5 * 44.0 32.8 15.6 19.6
Two 16.7 10.1 17.8 15.3 * 5.0 5.4 * 8.5 9.8 5.1 6.5
Three 8.9 6.0 11.6 9.7 * 5.0 2.7 * 0.9 0.0 2.7 1.4
Four 3.3 2.2 4.5 4.0 * 0.0 1.4 * 0.2 0.0 1.3 0.7
Five or more 1.0 1.0 2.5 2.0 * 0.0 0.7 * 0.0 0.0 0.1 0.7
Total 100.0 100.0 100.0 100.0 * 100.0 100.0 * 100.0 100.0 100.0 100.0
Number  of employment episodes                                           Income support spells of 12 to less than  24 months
None 44.4 48.6 39.7 45.6 * 82.8 79.1 * 45.0 48.3 69.0 62.9
One 20.0 18.8 17.1 17.0 * 10.3 6.5 * 41.5 30.3 18.1 21.0
Two 12.1 12.6 12.1 11.0 * 0.0 4.2 * 9.3 10.1 4.6 9.5
Three 9.2 7.5 11.9 9.0 * 3.4 4.2 * 3.1 6.7 3.5 0.0
Four 6.3 6.0 6.4 5.9 * 3.4 3.3 * 0.4 2.2 2.5 1.0
Five 5.4 3.2 5.5 4.7 * 0.0 0.7 * 0.5 1.1 1.3 1.9
Six 1.4 1.4 3.4 3.1 * 0.0 1.6 * 0.3 1.1 0.5 3.8
Seven or more 1.3 1.9 3.9 3.9 * 0.0 0.3 * 0.0 0.0 0.8 0.0
Total 100.0 100.0 100.0 100.0 * 100.0 100.0 * 100.0 100.0 100.0 100.0
Number  of employment episodes                                                  Income support spells of 24 months or more
None 46.7 43.4 42.0 46.4 * 84.6 76.6 * 44.0 44.4 67.1 55.2
One 16.8 21.7 14.1 14.0 * 7.7 4.7 * 34.3 40.7 19.3 20.9
Two 9.1 11.7 9.8 8.8 * 7.7 3.1 * 13.8 9.3 4.9 9.0
Three 7.7 6.2 7.5 6.8 * 0.0 2.3 * 5.8 1.9 3.0 4.5
Four 3.5 4.8 5.9 4.8 * 0.0 3.5 * 0.8 0.0 2.1 1.5
Five 6.0 2.4 5.2 4.4 * 0.0 3.1 * 0.9 0.0 1.2 3.0
Six 3.5 4.5 4.3 3.4 * 0.0 1.2 * 0.1 1.9 0.9 3.0
Seven 3.2 2.1 3.1 2.8 * 0.0 1.2 * 0.3 0.0 0.4 3.0
Eight 1.4 2.4 2.1 2.7 * 0.0 0.8 * 0.0 1.9 0.4 0.0
Nine or more 2.1 0.7 7.1 6.5 * 0.0 3.5 * 0.1 0.0 0.8 0.0
Total 100.0 100.0 100.0 100.0 * 100.0 100.0 * 100.0 100.0 100.0 100.0

* Sample  size too small for valid statistical analysis. Source: LDS 1% Sample.

Table 7: Time spent in employment for income support recipients who had an employment episode during an income support spell, LDS 1% Sample
  Unemployed (<21) Unemployed (>21) Unemployed (+60) Partner Allowance Parenting Single Parenting Partnered
Income support spells of less than  3 months
Mean duration of employment episodes (fortnights) 2.12 1.84 2.06 1.94 * * 2.12 * 2.94 2.59 2.42 2.44
Proportion of an income support spell spent  in employment 0.58 0.51 0.56 0.52 * * 0.61 * 0.84 0.70 0.73 0.71
Income support spells of 3 to less than  6 months
Mean duration of employment episodes (fortnights) 3.17 2.54 3.12 2.59 * * 4.10 * 6.98 4.77 4.35 4.03
Proportion of an income support spell spent  in employment 0.45 0.33 0.46 0.38 * * 0.60 * 0.73 0.53 0.56 0.51
Income support spells of 6 to less than  12 months
Mean duration of employment episodes (fortnights) 3.95 3.15 3.85 2.96 * * 3.53 * 10.35      10.19 5.77 5.45
Proportion of an income support spell spent  in employment 0.39 0.28 0.42 0.32 * * 0.39 * 0.64        0.61 0.47 0.41
Income support spells of 12 to less than  24 months
Mean duration of employment episodes (fortnights) 4.02 2.91 4.47 3.26 * * 6.35 * 16.01        8.40 7.72 5.40
Proportion of an income support spell spent  in employment 0.28 0.20 0.36 0.26 * * 0.46 * 0.58        0.40 0.42 0.30
Income support spells of 24 months or more
Mean duration of employment episodes (fortnights) 4.71 3.04 5.28 3.73 * * 6.25 * 23.59        17.63 11.65 12.43
Proportion of an income support spell spent  in employment 0.23 0.12 0.32 0.23 * * 0.42 * 0.51       0.40 0.33 0.40

Source: LDS 1% Sample.
* Sample  size too small for valid statistical analysis.

Consider  first the case of the distribution of employment episodes within short income support spells  (lasting less than three  months). As is evident in table  6, the vast majority of short spells do not include a single  employment episode. This is particularly the case for men; somewhat less so for women. Take first the case of Unemployed (21+) recipients. Short spells  including one or more employment episodes, comprise only 23.6  per cent  of all short spells  for male Unemployed (21+) recipients. Put in another  form, 76.4  per cent  of male Unemployed (21+) recipient short spells  do not include a single  episode of paid employment. In the case of female Unemployed (21+) recipients, 62.6  per cent  of short spells  do not include an episode of paid employment. This suggests that women are more likely  to have an employment episode in a short spell  than are men in this category.

The employment episode profile  of income support recipients in the Unemployed (<21) category, in the case of short spells, is very similar  to that for the Unemployed (21+) category. Here, 76.4  per cent  of men and 64.5  per cent  of women with  income support spells  of less than three  months  do not obtain  paid employment while on income support. In the case of the Partner Allowance and Unemployed (60+) categories, the proportion of recipients with income support spells  of less than three  months  duration who  experience an employment episode, is particularly low.

The employment episode experience of Parenting Single income support recipients is very different  (particularly for women) from the Unemployed (<21) and Unemployed (21+) groups. Among Parenting Single women, around  42 per cent  of all short spells  of income support include employment episodes, a figure  well  above comparable estimates for the unemployed. Likewise, the proportion of time spent  in paid employment while on income support for
female  Parenting Single  recipients is particularly high. Female Parenting Single recipients who  experienced an employment episode during  their  short spell  of income support spent, on average, 84 per cent  of their  time in paid employment. For Parenting Single men, the proportion experiencing an employment episode is roughly in line with  that for the Unemployed (21+) group. However, time spent  in employment while on income support is considerably higher  than for the unemployed group. The incidence of employment episodes among  Parenting Partnered recipients is very low. Only 16 per cent  of recipients in this category experience an employment episode within a short income support spell.

We now  turn to the employment episode experience of income support spells  of longer duration. It is the longer  duration income support spells  that are of greater interest from a policy perspective. This is because many unemployed income support recipients exit  the income support system  quickly by taking  up a longer-hours job with  relatively high earnings. Such individuals are immediately precluded from an income support payment and do not experience an employment episode (on the basis of the definition employed in this paper) in their  short income support spell. However, if the stepping-stone analogy prevails, the longer an unemployed recipient remains on income support without experiencing an employment episode, the lower will  be their  chances of exiting the income support system. Hence, our concern with  the issue  of the incidence of employment episodes among  the unemployed in longer  duration spells.

As is evident from tables  6 and 7, the longer  the income support spell  the more likely  it is that an episode of employment occurs. Likewise, the longer  the spell, the higher  the mean  duration of employment episodes. As the income support spell  duration increases, however, the proportion of time spent  in employment while on income support falls. These outcomes, which are evident across  the various  income support payment categories, are to be expected. However, what  is also evident is that the workforce experience patterns of the various  income support payment types  begins  to increasingly diverge as we move into income support spells of longer  duration.

Take first the case of the two main unemployed groups  of income support recipients; namely, the Unemployed (<21) and the Unemployed (21+) groups. The key point to arise from the evidence presented in table  6 is that increasing the length  of the income support spell  only marginally increases the incidence of employment episodes for such recipients. For example, among  very long duration income support spell  cases  (those  two years  or over), 46.4  per cent of male and 42 per cent  of female  Unemployed (21+) recipients had not experienced a single employment episode during  their  spell  on income support. The corresponding figures  for the Unemployed (<21) category are 43.4  per cent  (for men)  and 46.7  per cent  (for women).

While  the mean  duration of episodes of employment for the Unemployed (<21) and the Unemployed (21+) groups  increases the longer  the income support spell, the increase in the mean  duration is also relatively small. Among male Unemployed (21+) recipients, for example, the mean  duration of employment episodes for a spell  of less than three  months  is 1.94 fortnights. For income support spells  of three  months  to less than six months  the mean duration rises to only 2.59  fortnights while at the two years  or more income spell  point, the mean  duration rises to only 3.73  fortnights.

Consider  now  the case of the female  Parenting Single category. Here, the mean  duration of employment episodes rises sharply from 2.94  fortnights in the case of income support spells  of less than three  months  to 23.59  fortnights for spells  of 24 months  or more. Estimates of the proportion of time spent  in employment while on income support show  a similar  trend. As indicated in table  7, the female  Parenting Single recipient (who  gets a part-time  job while on income support) spends, on average, twice as much  time in employment as male Unemployed (21+) recipients.

Notice also that for the female  Parenting Single category, the number  of multiple employment episodes—or job churning—shows a much  more modest  rise as spell  length  increases, than for the male Unemployed (21+) category. For example, in the case of income support spells  of between three  and six months, close  to 20 per cent  of male Unemployed (21+) recipients experienced two or more employment episodes within their  spell. Among income support spells of six to less than twelve months, the job-churning rate rises to over 30 per cent. For spells  of twelve to less than 24 months, close  to 40 per cent  of male Unemployed (21+) recipients had more than one employment episode during  their  income support spell. Now consider the case of female  Parenting Single recipients. Only 3.9 per cent  of Parenting Single recipients with  income support spells  of between three  and six months ‘job churned’. For spells of 12 to less than 24 months, 13.6  per cent  of female  Parenting Single  recipients had more than >one employment episode during  their  spell. What this suggests is that there  is a relatively high degree of churning between jobs of short duration for a minority of unemployed recipients but a more stable  job profile  for the female  Parenting Single category. Consequently, when  female Parenting Single  recipients do get a job, they  stay in employment for a much  longer  time period.25

Having provided an in-depth  profile Having provided an in-depth  profile  of the Unemployed and Parenting Single groups, we can briefly  summarise the joint employment episode and income support spell  experience of the remaining income support recipient categories.

First, the incidence of employment episodes among  the Unemployed (60+) and Partner Allowance categories is very low compared with  the two main unemployed categories. The number  of male Unemployed (60+) recipients who  gained employment while on income support is so low that no meaningful estimates of time spent  in employment can be presented for this group. Second, the Parenting Partnered recipient category has a relatively low incidence of employment episodes across  all spell  lengths. However, the mean  duration of employment episodes and the proportion of time spent  in employment while on income support is relatively high compared to the unemployed categories suggesting that, in terms of part-time  job stability, this group  is similar  to the Parenting Single category.

Finally, notice  that women in each  recipient group  had longer  mean  durations of employment episodes and spent  a longer  proportion of their  income support spell  in employment than their male counterparts.

5.2 The employment and spell histories of 1995 recipients

To this point, our examination of the job and income support spell  histories of income support recipients has been  based  on the full set of spells  in the LDS 1% Sample. We can extend our understanding of the employment episode and spell  histories of the recipients in the LDS 1% Sample  by focusing on one particular group  of recipients (the  1995  group  is chosen) and following that group  over a discrete period of time. This ensures that we can, at least partially, control for group  compositional effects.

In table  8, we present the distribution of income support spells  for recipients who  are observed in 1995  in the LDS 1% Sample  (that  is, in the first year  of the data set). This cohort  is hereafter referred to as the 1995  cohort. The data in table  8 suggest a relatively high degree of income support churning across  the various  income support categories over the 1995  to 1999  period. The vast majority of recipients in the Unemployed (<21) and Unemployed (21+) categories experience more than one income support spell  over the period  1995  to 1999. For example, 32.7  per cent  of female  Unemployed (21+) recipients had two income support spells  over this period, 24.6  per cent  had three  income support spells, a further  13.6  per cent  had four spells  of income support, while 7.8 per cent  of recipients had five or more spells. Only 21.3  per cent  of the original 1995  cohort  of female  Unemployed (21+) recipients were restricted to a one- income support spell  over the 1995  to 1999  period. The incidence of single  income support spells  is more common  in the case of female  Partner Allowance recipients where 44.5  per cent of recipients experience a single  spell  of income support.

Table 8: Distribution of income support spells and employment episodes, 1995 cohort of income support recipients, LDS 1% Sample
  Women Men
Unemployed (<21) Unemployed (>21) Unemployed (+60) Partner Allowance Parenting Single Parenting Partnered Unemployed (<21) Unemployed (>21) Unemployed (+60) Partner Allowance Parenting Single Parenting Partnered
Distribution of income support spells
One income support spell 16.0 21.3 * 44.5 23.2 27.9 12.2 17.1 15.9 * 14.5 26.5
Two income support spells 29.3 32.7 * 36.0 36.2 35.0 23.8 26.9 54.5 * 30.1 37.7
Three  income support spells 30.7 24.6 * 15.2 23.3 20.5 28.1 24.7 18.2 * 29.0 22.3
Four income support spells 16.1 13.6 * 3.2 9.6 10.0 18.9 16.0 9.1 * 15.1 11.0
Five income support spells 5.4 5.8 * 0.4 4.5 4.8 9.0 9.7 0.0 * 7.0 2.3
Six income support spells  or more 2.4 2.0 * 0.8 3.3 1.9 8.0 5.7 2.3 * 4.3 0.3
Total 100.0 100.0 * 100.0 100.0 100.0 100.0 100.0 100.0 * 100.0 100.0
Distribution of employment episodes
No employment episodes 50.5 47.1 * 75.0 47.1 75.0 65.2 56.2 86.4 * 54.8 70.6
One employment episode 26.4 25.7 * 10.8 41.2 17.1 20.3 21.5 4.5 * 31.7 19.4
Two employment episode 11.3 11.8 * 4.2 8.3 3.5 7.1 9.7 4.5 * 8.1 6.5
Three  employment episode 5.8 6.6 * 3.0 2.5 1.8 3.4 5.2 2.3 * 3.2 1.3
Four employment episode 2.4 3.0 * 2.8 0.3 1.2 1.7 2.6 2.3 * 0.5 0.3
Five employment episode 2.0 2.2 * 1.1 0.4 0.5 0.7 1.7 0.0 * 0.5 0.3
Six employment episodes or more 1.7 3.6 * 3.0 0.2 1.0 1.6 3.1 0.0 * 1.1 1.6
Total 100.0 100.0 * 100.0 100.0 100.0 100.0 100.0 100.0 * 100.0 100.0

* Sample  size too small for valid statistical analysis. Source: LDS 1% Sample.

Table 8 also presents data on the distribution of employment episodes over the 1995  to 1999  period  for the 1995  cohort  of recipients. For men, the majority of recipients from the 1995  cohort  did not experience a single  employment episode (while on income support) at any time over the period  1995  to 1999. In the case of women, the absence of employment episodes is only marginally less striking. The incidence of ‘no employment episodes’ is greatest for the Partner Allowance and Unemployed (60+) categories where around three-quarters of income support recipients do not experience a single  episode of employment (while on income support) over the 1995 to 1999  period. It is lowest for Parenting Single recipients.

The ‘no employment episode’ experience is the modal employment episode outcome. The next  most typical outcome is a single  employment episode experience. Among Unemployed (21+) and Unemployed (<21) recipients, around  one-quarter experience a single  episode of employment over the 1995  to 1999  period. Multiple  employment episodes are also more frequent among  this group  of recipients.

We can obtain  a better  angle  on the incidence of multiple employment episodes over time by cross-tabulating income support spells  by employment episodes. In figures  23 to 32 we present the joint income support spell  and employment episode profile  of income support recipients. Each figure  displays the distribution of employment episodes by the number  of spells experienced over the 1995  to 1999  period. Among the Unemployed (<21) and Unemployed (21+) categories of recipients, the modal outcome is one of two/three income support spells and no employment episodes. In other  words, Unemployed (<21) and Unemployed (21+) recipients may typically be seen  two or three  times over a five year  time frame but not experience a single  episode of employment while on income support in these  various  spells of income support. Indeed, roughly half of all Unemployed (<21) and Unemployed (21+) recipients who  were observed in the LDS 1% Sample  in 1995, experience more than one income support spell  over the 1995  to 1999  period  income but do not experience an employment episode.

Interestingly, the distribution of employment episodes among  Unemployed (<21) and Unemployed (21+) recipients varies  little  as we move from a one-income support spell outcome to six or more income support spells  over the 1995  to 1999  period. In other  words, the relative incidence of job churning—defined here  in terms  of multiple employment episodes with  a given  period  of time rather  than within a given  income support spell—does not appear much  greater at the five or six income support spell  point as at the one or two income support spell  point. If anything, we can see more job churning among  one income support spell recipients as among  recipients with  multiple spells  of income support.

Figure 23: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (<21), women, 1995 to 1999, LDS 1% Sample

Figure 23: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (<21), women, 1995 to 1999, LDS 1% Sample

Source: LDS 1% Sample

Figure 24: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (<21), men, 1995 to 1999, LDS 1% Sample


Figure 24: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (<21), men, 1995 to 1999, LDS 1% Sample

Source: LDS 1% Sample

Figure 25: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (21+), women, 1995 to 1999, LDS 1% Sample

Figure 25: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (21+), women, 1995 to 1999, LDS 1% Sample

Figure 26: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (21+), men, 1995 to 1999, LDS 1% Sample

Figure 26: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (21+), men, 1995 to 1999, LDS 1% Sample

Figure 27: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (60+), men, 1995 to 1999, LDS 1% Sample

Figure 27: Employment  episodes by spells of income support,  1995 cohort,  Unemployed  (60+), men, 1995 to 1999, LDS 1% Sample

Figure 28: Employment  episodes by spells of income support,  1995 cohort,  Partner Allowance, women, 1995 to 1999, LDS 1% Sample

Figure 28: Employment  episodes by spells of income support,  1995 cohort,  Partner Allowance, women, 1995 to 1999, LDS 1% Sample

Figure 29: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Single, women, 1995 to 1999, LDS 1% Sample

Figure 29: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Single, women, 1995 to 1999, LDS 1% Sample

Figure 30: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Single, men, 1995 to 1999, LDS 1% Sample

Figure 30: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Single, men, 1995 to 1999, LDS 1% Sample

Figure 31: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Partnered, women, 1995 to 1999, LDS 1% Sample

Figure 31: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Partnered, women, 1995 to 1999, LDS 1% Sample

Figure 32: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Partnered, men, 1995 to 1999, LDS 1% Sample

Figure 32: Employment  episodes by spells of income support,  1995 cohort,  Parenting  Partnered, men, 1995 to 1999, LDS 1% Sample

Not surprisingly, the 1995  cohort  history  outcomes for male Unemployed (60+) and female Partner Allowance recipient categories are quite  different  to those  for the two main unemployed groups. The incidence of one and two income support spells  with  no employment episodes dominates the profile  of these  categories. It should  be noted  that the incidence of two income support spells  in the case of these  categories is artificially high as it includes transitions from the Mature Age Allowance and Partner Allowance to the Age Pension.

We now  turn to the experience of Parenting Single recipients. The histories profile  for female Parenting Single  recipients set out in Figure 29, reveals  the increased relevance of employment episodes among  this category of recipients. The incidence of one employment episode closely matches the incidence of no employment episodes across  the various  income support spell categories. As with  all other  income support categories examined, the incidence of joint income support and employment episode churning is very low.

5.3 Transition matrices

We close  our income support and employment episode histories analysis in this section by providing a snapshot of the income support and employment episode position of income support recipients at different  calendar time points  in the LDS 1% Sample. After choosing a cohort  of recipients at a particular point in time, we then examine their  income support and employment position at yearly intervals over the next  three  years. Of interest here  is the transition between income support payment categories over time (is there  much  movement between the various  income support categories?) and the transition between income support and non-support. The transition matrices may also shed light on the degree of persistence of part-time  employment and the coincidence of exit  from income support with  the receipt of part-time  labour  earnings. In the transition matrix  context, persistence in part-time  earnings occurs  when  earned income is received at both the beginning of the time period  and at the end of the period  (or at the point at the end of the last income support spell, whichever comes first). Our analysis in this context provides a natural  extension of the results  on the duration of employment episodes in the previous section. The coincidence of income support exit  with  the receipt of labour  income refers  to the question of whether, at the point of exit, a particular income support recipient was receiving labour  income.

The chosen starting  point date for our analysis is July 1996. Our interest lies, initially, in how the cohort  of income support recipients fared one year  later, that is, July 1997. Table 9 sets out transition matrices, by payment type  and by receipt of labour  income status, for men and women respectively. The top panel  of table  9 traces  out the income support payment location and earnings outcomes in 1997  for recipients in a particular payment type  category in July 1996 who  did not receive labour  income on that extract date. The second  panel  of table  9 refers  to those  recipients who  did receive labour  income in July 1996. The first set of columns provide data on income support payment type outcomes in 1997  while the second  set of columns provide  data on earnings status outcomes in 1997.

To interpret these  figures, consider the case of Unemployed (21+) recipients. Among male Unemployed (21+) recipients who  had zero labour  earnings in July 1996, 56.5  per cent  were located in the same income support payment category one year  later in July 1997. A further 35.7  per cent  were no longer  in receipt of an income support payment. The remaining male Unemployed (21+) recipients (totalling less than 10 per cent  of the cohort) were in other income support payment categories in July 1997. In the case of female  Unemployed (21+) recipients who  had zero earnings in 1996, 52.8  per cent  were in the Unemployed (21+) payment category in 1997  while 31.8  per cent  had exited the income support system.

Table 9: Transitions between income support payment categories by labour earnings status, 1996 to 1997, LDS 1% Sample
Payment category in 1996 Payment category in 1997 Earnings status in 1997
  Zero earnings at the end of the period Positive Earnings at the end of the period  
Unemployed (<21) Unemployed (>21) Unemployed (+60) Partner Allowance Parenting Single Parenting Partnered Other payments Income support exit Total Non-exit Exit Non-exit Exit Non-exit
Zero Earnings at the beginning of the period
Men
Unemployed (<21) 39.7 15.7 0.0 0.0 0.1 0.0 0.4 44.0 100.0 51.6 35.6 4.4 8.5 100.0
Unemployed (21+) 0.0 56.5 2.7 0.1 0.6 0.3 4.2 35.7 100.0 58.0 29.9 6.4 5.7 100.0
Unemployed (60+) 0.0 0.0 74.1 0.0 0.0 0.0 22.6 3.3 100.0 95.1 3.3 1.6 0.0 100.0
PartnerAllowance 0.0 4.4 15.6 68.9 0.0 0.0 6.7 4.4 100.0 86.7 4.4 8.9 0.0 100.0
ParentingSingle 0.0 11.7 0.0 0.0 63.5 2.5 5.6 16.8 100.0 76.1 13.2 7.1 3.6 100.0
ParentingPartnered 0.0 7.5 0.0 0.7 0.7 55.8 0.7 34.7 100.0 60.5 32.0 4.8 2.7 100.0
Total 5.0 43.5 7.9 0.6 2.8 1.8 5.1 33.2 100.0 61.0 27.8 5.8 5.4 100.0
Women
Unemployed (<21) 40.5 10.0 0.0 0.0 6.8 4.4 1.5 36.8 100.0 55.8 28.5 7.4 8.3 100.0
Unemployed (21+) 0.0 52.8 0.4 0.8 2.4 2.8 9.0 31.8 100.0 59.2 23.4 9.0 8.4 100.0
Unemployed (60+) * * * * * * * * * * * * * *
PartnerAllowance 0.0 0.9 0.4 66.8 0.1 0.3 21.1 10.4 100.0 87.9 10.4 1.7 0.0 100.0
ParentingSingle 0.0 2.4 0.0 0.0 78.8 5.2 4.7 9.0 100.0 80.5 7.6 10.5 1.3 100.0
ParentingPartnered 0.1 1.6 0.0 0.9 7.1 60.5 13.4 16.5 100.0 77.4 14.7 6.1 1.8 100.0
Total 2.8 13.4 0.1 7.4 29.2 19.5 9.5 18.1 100.0 74.1 14.7 7.8 3.3 100.0
Positive Earnings at the beginning of the period
Men
Unemployed (<21) 28.9 5.6 0.0 0.0 0.0 0.0 1.1 64.4 100.0 26.7 23.3 8.9 41.1 100.0
Unemployed (21+) 0.0 47.6 1.4 0.0 0.3 0.5 2.2 48.0 100.0 28.6 29.4 23.4 18.6 100.0
Unemployed (60+) 0.0 4.0 84.0 0.0 0.0 0.0 8.0 4.0 100.0 44.0 4.0 52.0 0.0 100.0
PartnerAllowance 0.0 9.1 0.0 54.5 0.0 0.0 9.1 27.3 100.0 18.2 9.1 54.5 18.2 100.0
ParentingSingle 0.0 2.3 0.0 0.0 75.0 0.0 6.8 15.9 100.0 27.3 9.1 56.8 6.8 100.0
ParentingPartnered 0.0 0.0 0.0 0.0 0.0 61.5 0.0 38.5 100.0 15.4 30.8 46.2 7.7 100.0
Total 2.7 39.4 3.3 0.6 3.6 1.2 2.5 46.6 100.0 28.5 27.0 24.9 19.5 100.0
Women
Unemployed (<21) 28.2 9.2 0.0 0.0 2.3 3.1 2.3 55.0 100.0 32.1 17.6 13.0 37.4 100.0
Unemployed (21+) 0.0 43.9 0.2 0.2 1.9 0.6 4.3 48.8 100.0 25.3 15.4 25.9 33.4 100.0
Unemployed (60+) * * * * * * * * * * * * * *
PartnerAllowance 0.0 1.1 0.0 58.2 0.0 0.0 13.2 27.5 100.0 35.2 16.5 37.4 11.0 100.0
Parenting Single 0.0 2.4 0.0 0.0 74.4 1.4 7.8 14.0 100.0 25.9 4.4 60.1 9.6 100.0
ParentingPartnered 0.0 0.0 0.0 0.9 5.6 57.0 15.9 20.6 100.0 36.4 4.7.8 43.0 15.9 100.0
Total 2.2 14.4 0.1 3.3 39.7 4.8 7.2 28.2 100.0 27.4 9.2 44.4 19.0 100.0

*Sample size too small for valid statistical analysis.
Source: LDS 1% Sample.

Turning  now  to the case of the male Unemployed (21+) recipients in receipt of labour earnings in July 1996, some 47.6  per cent  were in the Unemployed (21+) category in July 1997, while 48.0  per cent  had exited the income support system. The respective figures  for female  Unemployed (21+) recipients was 43.9  per cent  (same  payment category) and 48.8  per cent  (exit).

Consider  now  the final columns of table  9. These columns present information on the earning status position of income support recipients at the end of the year  (July 1997). Among male Unemployed (21+) non-earners in July 1996, 58.0  per cent  were non-earners receiving an income support payment of one form or another  at the end of the period. A further  29.9  per cent  exited the income support system  as a non-earner. Some 5.7 per cent  of male Unemployed (21+) non-earners in July 1996  exited the income support system  while earning with  a further 6.4 per cent  earning at the end of the period  but not exiting.

As reflected in table  9, the majority of income support recipients are located in the same payment category at the end of the period  as at the beginning of the period. This can be easily verified  by tracing down  the diagonal elements on the payment type  matrix. One exception to this general rule occurs  in the case of the Unemployed (<21) category. In the case of this group, exit  rates are relatively high and, in addition, there  is a degree of movement from the Unemployed (<21) category to the Unemployed (21+) category. This progression from the Unemployed (<21) group  to the Unemployed (21+) category for long-term  unemployed young people is obviously a natural  age-related phenomenon.

The one-year  transition analysis reveals  a much  higher  rate of one-year  exit  for those  with labour  earnings at the beginning of the period  than those  without labour  earnings at this point. Among the pool of income support recipients being  considered in our analysis, 28.2  per cent  of female  recipients with  labour  income at July 1996  were no longer  on income support at July 1997. This compares with  an exit  rate of 18.1  per cent  for those  female  recipients without labour  income. Among men, 46.6  per cent  of recipients with  labour  income at July 1996  were no longer  on income support at July 1997  compared with  an exit  rate figure  of 33.2  per cent for those  recipients without labour  income

Not only is the one year  exit  rate considerably higher  at the aggregate level  for those  with positive labour  income, it is also higher  for each individual income support payment category. For example, while the exit  rate for female  Parenting Single income support recipients with no labour  earnings at July 1996  is 9.0 per cent, in the case of those  with  positive labour earnings at July 1996, the exit  rate increases to 14.0  per cent. The exit rate for unemployed recipients with  positive labour  earnings is also considerably higher  than that for non-earners. The role of labour  earnings in positively influencing the exit  rate is considered in greater detail in the following chapter.

A high degree of part-time earnings persistence is also evident among  the selected payment categories in the LDS 1% Sample. Among those  in receipt of labour  earnings at July 1996, 44.4  per cent  of women and 24.9  per cent  of men were both still on income support and in receipt of labour  income at the end of the period. To this group  can be added  those  earners who  exited the income support system  while in receipt of labour  earnings. Among men, 19.5 per cent of earners at the beginning of the period exited while earning. The corresponding estimate for women is 19.5 per cent. Taken together more than half of those with income support at the beginning of the period either exited while still receiving labour income or were in receipt of part-time earnings at the end of the period. In short, part-time employment persistence is a more prominent phenomenon than stepping-stone exit.

The level of part-time persistence is much higher among women than men. Nearly half of all women receiving labour income at the beginning of the period were still on income support at the end of the period and still in receipt of part-time earnings. The higher rate of part-time employment persistence found in the LDS 1% Sample for female income support recipients is consistent with the higher levels of voluntary part-time employment among women in the Australian labour market.

Our transition matrix analysis is extended into a second year in table 10. The third-year transition matrixes are displayed in table 11. As we move into the second and third years of the transitions analysis, the proportion of income support recipients who are no longer on an income support payment rises.However, the increase in the proportion of recipients who exit the income support system in 1998 and 1999 is relatively small. Among men, the one-year exit rate for earners was 46.6 per cent. In the second year, the exit rate had increased to 51.4 per cent but in the third year the exit rate had only increased to 56.5 per cent. Among non-earners, the exit rate for men was 33.2 per cent in the first year. This had risen to only 35.7 per cent in the second year and 40.5 per cent in the third year. A similar profile of slowly rising exit rates over the three-year period is also evident for women.26

Table 10: Transitions between income support payment categories by labour earnings status, 1996 to 1998, LDS 1% Sample
Payment category in 1996 Payment category in 1998 Earnings status in 1998
  Zero earnings at the end of the period Positive Earnings at the end of the period  
Unemployed (<21) Unemployed (>21) Unemployed (+60) Partner Allowance Parenting Single Parenting Partnered Other payments Income support exit Total Non-exit Exit Non-exit Exit Non-exit
Zero Earnings at the beginning of the period
Men
Unemployed (<21) 23.8 29.4 0.1 0.0 0.0 0.0 1.5 45.2 100.0 51.7 37.0 3.1 8.2 100.0
Unemployed (21+) 0.3 48.8 3.2 0.1 0.6 0.5 8.3 38.0 100.0 56.6 32.1 5.3 5.9 100.0
Unemployed (60+) 0.0 0.0 47.5 0.0 0.0 0.0 46.6 6.1 100.0 92.0 6.1 1.9 0.0 100.0
PartnerAllowance 0.0 4.4 20.0 46.7 0.0 0.0 17.8 11.1 100.0 82.2 11.1 6.7 0.0 100.0
ParentingSingle 0.0 17.8 0.0 0.0 54.3 2.0 9.1 16.8 100.0 76.1 13.7 7.1 3.0 100.0
ParentingPartnered 0.0 12.2 0.0 1.4 2.0 33.3 2.7 48.3 100.0 43.5 46.3 8.2 2.0 100.0
Total 3.3 40.0 6.2 0.5 2.5 1.4 10.4 35.7 100.0 59.4 30.3 4.9 5.4 100.0
Women
Unemployed (<21) 25.9 17.9 0.0 0.0 12.4 7.8 1.7 34.2 100.0 60.3 25.5 5.4 8.7 100.0
Unemployed (21+) 0.4 42.1 0.3 1.5 4.4 3.2 14.8 33.2 100.0 59.6 23.9 7.2 9.3 100.0
Unemployed (60+) * * * * * * * * * * * * * *
PartnerAllowance 0.0 1.3 0.3 53.0 0.1 0.3 29.3 15.6 100.0 82.1 15.3 2.3 0.3 100.0
ParentingSingle 0.0 4.4 0.0 0.0 71.6 6.5 7.2 10.3 100.0 77.1 8.0 12.6 2.3 100.0
ParentingPartnered 0.0 2.5 0.0 1.7 12.0 48.1 14.7 21.0 100.0 73.3 18.5 5.7 2.5 100.0
Total 1.9 12.6 0.1 6.3 29.0 16.8 12.9 20.4 100.0 71.6 16.3 7.9 4.1 100.0
Positive earnings at the beginning of the period
Men
Unemployed (<21) 11.1 17.8 0.0 0.0 0.0 0.0 2.2 68.9 100.0 27.8 32.2 3.3 36.7 100.0
Unemployed (21+) 0.3 39.0 2.3 0.0 0.5 0.6 4.6 52.6 100.0 31.1 33.4 16.3 19.3 100.0
Unemployed (60+) 0.0 0.0 64.0 0.0 0.0 0.0 32.0 4.0 100.0 68.0 4.0 28.0 0.0 100.0
Partner  Allowance 0.0 9.1 0.0 18.2 0.0 0.0 9.1 63.6 100.0 18.2 45.5 18.2 18.2 100.0
Parenting Single 0.0 2.3 0.0 0.0 65.9 0.0 15.9 15.9 100.0 50.0 4.5 34.1 11.4 100.0
Parenting Partnered 0.0 7.7 0.0 0.0 0.0 38.5 0.0 53.8 100.0 0.0 46.2 46.2 7.7 100.0
Total 1.2 33.6 3.5 0.2 3.4 1.0 5.6 51.4 100.0 32.0 31.5 16.6 19.9 100.0
Women
Unemployed (<21) 19.8 19.8 0.0 0.0 5.3 3.8 0.8 50.4 100.0 38.9 19.8 10.7 30.5 100.0
Unemployed (21+) 1.1 36.0 0.2 0.6 3.6 0.4 7.5 50.5 100.0 31.5 18.2 18.0 32.3 100.0
Unemployed (60+) * * * * * * * * * * * * * *
Partner  Allowance 0.0 1.1 0.0 45.1 0.0 0.0 23.1 30.8 100.0 48.4 20.9 20.9 9.9 100.0
Parenting Single 0.0 3.6 0.0 0.0 61.4 3.1 10.8 21.1 100.0 32.9 7.8 45.9 13.3 100.0
Parenting Partnered 0.0 0.9 0.0 0.0 5.6 45.8 16.8 30.8 100.0 47.7 14.0 21.5 16.8 100.0
Total 1.9 13.7 0.1 2.7 33.7 5.0 10.1 32.9 100.0 34.8 12.8 32.3 20.1 100.0

 

Table 11: Transitions between income support payment categories by labour earnings status, 1996 to 1999, LDS 1% Sample
Payment category in 1996 Payment category in 1998 Earnings status in 1998
  Zero earnings at the end of the period Positive Earnings at the end of the period  
Unemployed (<21) Unemployed (>21) Unemployed (+60) Partner Allowance Parenting Single Parenting Partnered Other payments Income support exit Total Non-exit Exit Non-exit Exit Non-exit
Zero Earnings at the beginning of the period
Men
Unemployed (<21) 13.1 34.3 0.0 0.0 0.3 0.4 2.5 49.4 100.0 47.5 40.4 3.1 9.0 100.0
Unemployed (21+) 0.2 41.1 3.6 0.1 0.8 0.6 10.6 43.1 100.0 52.6 36.8 4.3 6.3 100.0
Unemployed (60+) 0.0 0.0 25.2 0.0 0.0 0.0 66.8 8.0 100.0 91.3 8.0 0.7 0.0 100.0
PartnerAllowance 0.0 4.4 17.8 31.3 0.0 0.0 26.7 20.0 100.0 80.0 17.8 0.0 2.2 100.0
ParentingSingle 0.0 16.8 0.0 0.0 47.4 1.5 11.7 22.3 100.0 72.1 18.3 5.6 4.1 100.0
ParentingPartnered 0.0 11.6 0.0 2.0 2.7 23.1 4.8 55.8 100.0 37.4 52.4 6.8 3.4 100.0
Total 1.8 35.0 4.7 0.4 2.4 1.1 14.0 40.5 100.0 55.5 34.6 4.9 6.0 100.0
Women
Unemployed (<21) 13.5 21.6 0.0 0.0 15.3 7.2 3.5 39.0 100.0 54.0 29.2 7.0 9.8 100.0
Unemployed (21+) 0.4 33.1 0.2 1.5 6.1 3.2 18.5 36.9 100.0 57.8 25.9 5.3 11.1 100.0
Unemployed (60+) * * * * * * * * * * * * * *
PartnerAllowance 0.0 0.9 0.1 46.1 0.1 0.1 40.8 16.3 100.0 82.2 15.9 1.4 0.4 100.0
ParentingSingle 0.0 5.2 0.0 0.0 64.3 6.8 9.7 13.9 100.0 73.0 11.5 13.1 2.4 100.0
ParentingPartnered 0.0 3.2 0.0 2.0 13.8 40.9 17.7 22.4 100.0 70.7 20.0 6.9 2.4 100.0
Total 1.0 11.4 0.1 5.2 27.6 18.4 16.6 23.2 100.0 68.7 18.6 8.1 4.6 100.0
Positive earnings at the beginning of the period
Men
Unemployed (<21) 6.7 20.0 0.0 0.0 0.0 1.1 2.2 70.0 100.0 24.4 35.6 5.6 34.4 100.0
Unemployed (21+) 0.4 30.9 2.8 0.1 0.5 1.2 6.0 58.0 100.0 30.7 38.6 11.3 19.4 100.0
Unemployed (60+) 0.0 0.0 48.0 0.0 0.0 0.0 48.0 4.0 100.0 84.0 4.0 12.0 0.0 100.0
PartnerAllowance 0.0 18.2 0.0 9.1 0.0 0.0 27.3 45.5 100.0 36.4 18.2 18.2 27.3 100.0
ParentingSingle 0.0 2.3 0.0 0.0 50.0 0.0 15.9 31.8 100.0 38.6 20.5 29.5 11.4 100.0
ParentingPartnered 0.0 7.7 0.0 0.0 0.0 23.1 0.0 69.2 100.0 15.4 53.8 15.4 15.4 100.0
Total 0.9 27.3 3.5 0.2 2.7 1.4 7.4 56.5 100.0 31.7 36.6 11.7 20.0 100.0
Women
Unemployed (<21) 6.9 13.7 0.0 0.0 12.2 2.3 1.5 63.4 100.0 28.2 32.8 8.4 30.5 100.0
Unemployed (21+) 0.4 24.6 0.4 0.9 4.1 1.3 11.6 56.7 100.0 29.8 24.6 13.5 32.1 100.0
Unemployed (60+) * * * * * * * * * * * * * *
PartnerAllowance 0.0 1.1 0.0 36.3 0.0 0.0 26.4 36.3 100.0 50.5 24.2 13.2 12.1 100.0
ParentingSingle 0.0 3.2 0.0 0.0 53.6 3.0 11.3 28.9 100.0 34.1 12.9 37.0 16.0 100.0
ParentingPartnered 0.0 0.9 0.0 0.0 7.5 30.8 27.1 33.6 100.0 51.4 16.8 15.0 16.8 100.0
Total 0.7 9.8 0.1 2.2 30.5 4.1 12.4 40.2 100.0 34.4 18.6 25.4 21.5 100.0

*Sample size too small for valid statistical analysis.
Source: LDS 1% Sample.

5.4 Summary

As with  other  results  in this paper, significant differences can be found between the various income support categories in the income support spell  and employment history  profiles  of recipients. These profiles  are obtained by tracking individual income support recipients through the various  fortnightly records that make up the LDS 1% Sample. Our employment episode and income support histories analysis suggests that the Unemployed (<21) and Unemployed (21+) recipient groups  typically exhibit both shorter  income support spells  and shorter  employment episodes than can be found for each  of the remaining categories. Moreover, as the income support spell  lengthens, the gap between the mean  duration of employment episodes for Parenting Single  recipients and those  for the Unemployed (<21) and Unemployed (21+) groups increases. The female  Parenting Single income support recipient, who  obtains  work while on income support, spends, on average, twice as much  time in employment as female Unemployed (21+) recipients who  obtain  employment while on income support.

Likewise, there  is a tendency for the Unemployed (<21) and Unemployed (21+) groups  to experience a somewhat higher  degree of ‘job churning’ (or multiple employment episodes) than can be found among  Parenting Single recipients. In other  words, Unemployed (<21) and Unemployed (21+) income support recipients are more likely  than their  Parenting Single counterparts to move between different  jobs while on income support. In short, the Parenting Single  pattern of engagement with  the labour  market  is much  more one of long-duration spells of income support and stable  patterns of income receipt than is evident for the two main unemployed recipient groups.

A point of concern to policy-makers from our results  is the relatively low incidence of take-up of part-time  employment positions among  income support recipients with  medium to long- duration spells  of income support. To take one example, among  male Unemployed (21+) recipients with  income support spells  lasting  more than two years, 46 per cent  of recipients had not experienced a single  employment episode during  their  spell  on income support. This is of significant policy concern as the longer  the unemployed person  remains away  from any sort of employment, the more their  skills and, for some, their  mental  health  will  deteriorate. A focus of any employment-related strategy for the long-term  unemployed, given  this evidence, is the provision of wage  subsidies to enable such recipients to engage in the formal labour  market.

[ Return to Top   Return to Section ]

6 Labour earnings and exit from the income support system

The most important policy issue  we address  in this paper  is the question of whether obtaining short-hour  jobs improves the chances of final exit  from income support. In this chapter we first examine trends  in exit  rates for our various  classes  of income support recipients over time. The exit  rate is defined  as the proportion of income support recipients at time period  t who are not on income support at time period  t+1. We conclude our review of the impact  of earnings on the chances of exit  from the income support system  with  a hazard  analysis of exit from income support. Empirical  hazard  functions are presented and the proportional hazards model  is estimated across  various  income support payment classes  to determine the role earnings play in exit  from spells  of income support.

6.1 Time series exit rates

We define  the time series  exit  rate as the proportion of income support recipients at time t who  were no longer  on income support at time t+1. An exit  rate of 30 per cent, for example, indicates that 30 per cent  of recipients at a given  point in time will  no longer  be on income support in the next  time period. Two time series  exit  rate series  are estimated for the 1995 to 1999  period. The first is a short-run exit  rate. Here, an income support payment recipient exits  the income support system  if she or he is not observed receiving an income support payment more than two fortnights later. The second  exit  rate series  is a long-run series. A long-run exit  is said to occur  if the income support recipient is no longer  receiving income support one year  from the date of observation. Given the greater time period  covered by the long-run exit  rate measure, we would  expect the long-run exit  rate to be higher  than the short-run exit  rate.

The time series  profile  of both short-run and long-run exit  rates is displayed for the various income support payment types  in figures  33 to 44. Our main point of focus lies with  whether exit  rates are higher  for income support recipients in paid employment (while on income support) than for those  not receiving earned income. As is clear  from both the short-run and long-run exit  rate trends, the gross exit  rate for those  with  labour  earnings is higher  than for comparable income support recipients without labour  earnings. Put in another  form, those recipients who  are in paid employment positions at one point in time have a higher  chance than non-earners of exiting the income support system  in the near future. This result  applies across  the various  income support payment types  and for both women and men.

Figure 33: Short-run  exit rates for Unemployed  (<21) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 33: Short-run exit rates for Unemployed (<21) income support payment recipients, moving average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 34: Long-run exit rates for Unemployed  (<21) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 34: Long-run exit rates for Unemployed  (<21) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 35: Short-run  exit rates for Unemployed  (21+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 34: Long-run exit rates for Unemployed  (<21) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 36: Long-run exit rates for Unemployed  (21+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 36: Long-run exit rates for Unemployed  (21+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 37: Short-run  exit rates for Unemployed  (60+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 37: Short-run  exit rates for Unemployed  (60+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 38: Long-run exit rates for Unemployed  (60+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 38: Long-run exit rates for Unemployed  (60+) income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 39: Short-run  exit rates for Partner Allowance  income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 39: Short-run  exit rates for Partner Allowance  income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 40: Long-run exit rates for Partner Allowance  income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 40: Long-run exit rates for Partner Allowance  income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 41: Short-run  exit rates for Parenting  Single income support  payment  recipients, moving average estimates, LDS 1% Sample

Figure 41: Short-run  exit rates for Parenting  Single income support  payment  recipients, moving average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 42: Long-run exit rates for Parenting  Single income support  payment  recipients, moving average estimates, LDS 1% Sample

Figure 42: Long-run exit rates for Parenting  Single income support  payment  recipients, moving average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 43: Short-run  exit rates for Parenting  Partnered income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 43: Short-run  exit rates for Parenting  Partnered income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

Figure 44: Long-run exit rates for Parenting  Partnered income support  payment  recipients, moving  average estimates, LDS 1% Sample

Figure 44: Long-run exit rates for Parenting  Partnered income support  payment  recipients, moving  average estimates, LDS 1% Sample

Source: LDS 1% Sample

The difference between exit  rates for earners and non-earners is greatest for the Unemployed(<21) and Unemployed (21+) groups. The gap in exit  rates for the two groups  is around  6 to 10 per cent  in the case of the short-run exit  rate and up to 15 per cent  in the case of the long- run exit  rate.

Among male Unemployed (<21) recipients, the short-run exit  rate for earners averages around 24 per cent  over the 1995  to 1999  time period.27 The corresponding figure  for non-earners is 15 per cent. In the case of the long-run exit  rate, the average long-run exit  rate for male Unemployed (<21) recipients who  are in paid employment positions is 56 per cent  compared to an average of 41 per cent  for non-earners. For women, the average short-run exit  rate for earners is around  19 per cent, appreciably lower than the corresponding male average of 24 per cent. For non-earners, the short-run exit  rate averages around  12 per cent. The average long-run exit  rate for female  Unemployed (21+) recipients who  are in paid employment positions is 53 per cent  compared to an average of 35 per cent  for non-earners.

Exit rates for Unemployed (21+) recipients are somewhat lower than for Unemployed (<21) recipients. However, the gap between earner and non-earner exit  rates remains. Among male Unemployed (21+) recipients, the short-run exit  rate for earners averages around  17 per cent over the 1995  to 1999  time period. The corresponding estimate for non-earners is 10 per cent. The gap between earners and non-earners for male Unemployed (21+) recipients is even greater in the case of the long-run exit  rate. The average long-run exit  rate for male Unemployed (21+) recipients who  are in paid employment positions is 45 per cent  compared to an average of 33 per cent  for non-earners.

Among female  Unemployed (21+) recipients, the average short-run exit  rate for earners is around  15 per cent, a little  lower than the corresponding male average. For non-earners, the short-run exit  rate averages around  9 per cent. As with  men, the gap between earners and non- earners for female  Unemployed (21+) recipients is even  greater in the case of the long-run exit rate. The average long-run exit  rate for female  Unemployed (21+) recipients who  are in paid employment positions is 44 per cent  compared to an average of 30 per cent  for non-earners.

Exit rates for the remaining unemployed income support payment categories are considerably lower than for the Unemployed (<21) and Unemployed (21+) categories. Likewise, the gap between the earner and non-earner recipient is generally much  smaller. In the case of the male Unemployed (60+) category, both short-run and long-run exit  rates are very low with  no discernible difference found between earner and non-earner exit  rates. For female  Partner Allowance recipients, short-run exit  rates for earners average around  4 per cent  with  only a 1 to 2 per cent  differential evident in respect of non-earners. Interestingly, the differential between the earner and non-earner exit  rate in terms  of the long-run exit  rate is much  higher and is closer  to 9 per cent.

Turning  now  to income support recipients in the families  with  children income support program, the average short-run exit  rate for female  Parenting Single  earners is around  3 per cent. This figure  is well  below that recorded for the main unemployed income support recipient categories. The average short-run exit  rate for female  Parenting Single recipients lies only 1 per cent  above the corresponding rate for non-earners. The differential between the earner and non-earner exit  rate is, however, higher  for female  Parenting Single  recipients in the case of the long-run exit. Over the 1995  to 1999  period, the long-run exit  rate for earners averaged around  the 13 per cent  level  while for non-earners the exit  rate averaged around 7 per cent. The male Parenting Single exit  rates exhibit a deal of instability over the 1995  to1999  period  although it must be remembered that the number  of male Parenting Single recipients in the Longitudinal Data Set (LDS) 1% Sample  is relatively small (averaging only around  250 recipients at any point in time). The male Parenting Single short-run exit  rate averages around  5 per cent  for earners and 3 per cent  for non-earners. The corresponding long-run exit  rates are 19 per cent  for earners and 7 per cent  for non-earners.

In the case of female  Parenting Partnered income support recipients, the short-run exit  rate lies somewhere between the relatively high exit  rates for the Unemployed (<21) and Unemployed (21+) recipient groups  and the relatively low exit  rates for Parenting Single recipients. Over the period  1995  to 1999, the average short-run exit  rate for female  Parenting Partnered recipients was 8 per cent  for earners and 5 per cent  for non-earners. The long-run exit  rate for female  Parenting Partnered recipients who  were in receipt of earned income was 23 per cent compared with  14 per cent  for non-earners. Male Parenting Partnered exit  rates are somewhat higher  than the corresponding rates for women although the gap between the earner and non- earner exit  rates is very small.

The analysis of exit  rates between the various  income support payment categories has revealed a significant gender divide. Across all payment categories considered, male exit  rates lie well above the corresponding rates for women.

Our final point of interest lies in how  exit  rates have moved  over time. For the unemployed income support payment categories, a slow  decline over the initial  years  of the LDS 1% Sample is evident in terms  of short-run and long-run exit  rates. Exit rates rose slightly in the final year  of the LDS 1% Sample. This trend  is broadly  consistent with  the facts of the labour  market  during this time; namely, a sluggish performance in terms  of unemployment through 1995  to 1997  and a strong  improvement in the Australian  labour  market  through 1998  and 1999. In the case of Parenting Single income support recipients, little  by way  of a time trend  is evident. Exit rates remain  stable  in spite  of broader  movements in the labour  market.

6.2 Empirical hazard functions

In the final stage  of the analysis, we examine the question of exits  from the perspective of what is known as hazard  analysis. Here the focus is not on the proportion of income support recipients who  exit  the income support system  at a particular calendar time point but on exit from income support at various  points  within an income support spell. The hazard  rate for a given  fortnight within an income support spell  is derived by dividing the number  of recipients in a chosen income support category who  were observed to exit  from income support in that fortnight by the number  of recipients who  potentially could  have exited in that fortnight. The number  who  potentially could  have exited is given  by the number  of recipients in the previous fortnight less the number  of recipients censored during  that fortnight. An income support spell is right censored if the spell  remains uncompleted at the end of the LDS 1% Sample period. For the purposes of our hazard  rate analysis we only include fresh spells  of income support in the LDS 1% Sample. Existing spells  in the LDS 1% Sample  in June 1995  are deleted.

The empirical hazard  function  represents the plot of the hazard  rate beginning with  the first fortnight and ending  with  the 100th  fortnight.28 To interpret the graph, if, for example, the hazard rate is 0.08  at fortnight one, that tells us that 8 per cent  of income support recipients who potentially could  have left the income support system  in that particular fortnight actually did so in that fortnight. If the hazard  rate falls to 0.06  in the following fortnight, that means  that the conditional exit  rate in that fortnight has fallen to 6 per cent. The higher  the hazard  rate, the greater the conditional rate of exit  while the lower the hazard  rate the lower the conditional
rate of exit. If the hazard  function  shows  a positive slope  over a particular range, then the hazard  is said to be rising, meaning that the ‘risk’ of exit  from the income support system  is increasing. Conversely, a negative slope  over a particular range  indicates a falling hazard, or a declining ‘risk’ of exit  from the income support system  is increasing. Hazard function  plots for the unemployed have traditionally revealed a declining hazard  or increasing duration dependence, suggesting that it becomes more difficult  to leave  the state of unemployment the longer  the duration of unemployment. Figure 45: Hazard function for Unemployed  (21+) income support  payment  recipients

Figure 46: Hazard function for parenting income support  payment  recipients, LDS 1% Sample

Figure 46: Hazard function for parenting income support  payment  recipients, LDS 1% Sample

Source: LDS 1% Sample

Plots of the empirical hazard  function  for the male and female  Unemployed (21+) and the female  Parenting Single and the Parenting Partnered categories are given  in figures  45 and 46. For both women and men in the Unemployed (21+) category of income support recipients, an increasing hazard  is evident in the first two months  on income support. In the case of male Unemployed (21+) recipients, the hazard  rate rises from 0.048  in the first fortnight to 0.055  in the 8th fortnight. For women, the hazard  rate rises more rapidly going  from 0.044  in the first fortnight to 0.075  in the 4th fortnight. From that point, there  is a sharp  decline in the hazard rate, followed by a gradual  slow  decline thereafter. The male and female  hazard  plots show  a roughly similar  trend. A high degree of volatility in the hazard  rate is evident above the 85th fortnight point.

The hazard  rate follows  a different  path for the two parenting income support payment categories. Here the hazard  rate declines continuously from the first fortnight, but at a relatively slow  rate. For female  Parenting Single recipients, the hazard  rate begins  at 0.031  falling to around  0.024  in the 8th fortnight. The hazard  rate remains below the corresponding hazard
plot for female  Unemployed (21+) recipients throughout all fortnightly intervals. In the case of female  Parenting Partnered recipients, the shape  of the plot of the hazard  rate closely resembles that for female  Parenting Single  recipients in that there  is evidence of a declining hazard  throughout. However, the hazard  rate is well  above that for Parenting Single recipients in the early  fortnights of the income support spell.

6.3 Estimated proportional hazard functions29

As noted  above, the hazard  function  gives the probability that an individual leaves  a particular state at a particular point in time conditional on being  in that state prior to that point. In the present context, the hazard  function  gives the probability of exiting income support at a given point in time, given  that individual had been  on income support up to that point. We now move on to the estimation of a statistical model  that attempts to explain movements in the hazard  rate.

For this purpose we utilize  Cox’s  popular proportional hazards  model. In this model, we estimate the hazard  function  using  the following functional form:

h (t, X) = h (t,0)  exp(   ’X)

where h (t,0)  is the baseline hazard  rate and X represents a set of determinants we believe may influence the hazard  rate. Notice that the baseline hazard  rate reflects the influence of spell duration on the hazard  rate and is independent of the set of X determinants. Likewise, the set
of X determinants have a constant effect on the hazard  rate over time.

We estimate the proportional hazard  function  for the Unemployed (21+) and female  parenting payment categories of income support recipients. Explanatory variables incorporated in the analysis are similar  to the set used in the logit model  of earned income receipt in Chapter 5 but are repeated here  for convenience.

Socio-demographic variables include:

  • Number of dependent children and their  age structure (Number of dependent children, Aged 0–6, Aged 7–12, Aged 13–15, Aged over 15);
  • Country  of birth (Main English speaking country, Non  English speaking country; the default  is Australian  born);
  • Age (Age and Age squared);
  • Aboriginal  and Torres Strait Islander  status (Aboriginal and Torres Strait Islander status);
  • Housing tenure (Home purchaser, Other home owner, Private rent, Government rent, Other rent; the default  is outright home owner).

Income  support arrangement variables include:

  • Rent ($)(the amount  of rent assistance received by the income support recipient);
  • Rent assistance sharer  status  (Sharer);
  • Average unearned income;
  • Average part-time earnings during the spell;
  • Partner receives an unemployment-related benefit;
  • Average job search activity;
  • Duration on income support > 30 days;
  • Repeat spells on income support during sample period.

The estimated hazard  function  results  are contained in table  12 in the case of Unemployed (21+) recipients and table 13 in the case of Parenting Payment recipients.

As can be seen  from tables  12 and 13, older  income support recipients have a lower hazard rate, meaning that such recipients have a lower conditional probability of exiting the income support system  than do younger recipients controlling for all other  factors. The effect of age on the hazard  rate is non-linear; the decline in the hazard  rate slows  down  as recipients get older. The age-hazard  rate relationship applies across  all of groups  considered in the analysis.

Country  of birth does not provide  a consistent effect on the hazard  rate across  the various income support categories. For male Unemployed (21+) recipients, the hazard  rate increases significantly for those  recipients born in the main English-speaking countries relative to the Australian  born. The risk ratio for male Unemployed (21+) recipients born in the main English- speaking countries is 1.13  (p=0.00). This indicates that the hazard  rate is 13 per cent  above that for the Australian  born all other  things  being  equal. A much  smaller  effect is found in the case of female  Unemployed (21+) recipients and the effect is barely  significant (p=0.10). There is no significant effect of this variable in the case of female  Parenting Single  recipients but a significant positive effect is found in terms  of female  Parenting Partnered recipients. For this category of recipients, a negative effect on the hazard  rate is found for those  born in the non- English-speaking countries. This means  such recipients have a lower probability of exit  from the income support system  than do those  born in Australia. Aboriginal  and Torres Strait Islander male Unemployed (21+) income support recipients have significantly lower hazard  rates than non-Aboriginal and Torres Strait Islanders. No significant effect is found in the case of the other income support recipient categories.

A number  of housing-related effects  are evident in the results. For male Unemployed (21+) recipients, home purchasers tend to have a significantly higher  probability of exit  from the income support system  while renting in the Government sector  significantly lowers the risk ratio. A significantly lower risk ratio for those  recipients renting in the Government sector  is also evident in the case of female  Unemployed (21+) recipients and female  Parenting Partnered recipients. We noted  in Chapter 4 that government rental  status may reflect  long-term socio-economic disadvantage. Other housing-related effects  of interest include the fact that an increase in rent assistance payments lowers the risk ratio (or alternatively, lowers the probability of exit  from the income support payment system) as does being  a ‘rent  sharer’. A ‘rent  sharer’ refers  to the fact that the recipient is in shared  accommodation, is not married and is subject to sharer  legislative provisions with  regard  to rent assistance. The effect of young  dependent children is to reduce the hazard  rate.

We now  turn to the effect of labour  earnings on the hazard  rate. Across the various  income support categories, higher  levels  of earned income increases the hazard  rate. In other  words, higher  levels  of earned income increases the probability of exit  from the income support system. This result  is perfectly consistent with  the results  presented in the previous section in respect to the time series  exit  rate and also our results  in the previous chapter on transitions. Higher levels  of unearned income also have this effect.

Two final effects  are worth  emphasising. First, the exceptionally strong  positive effect of the job search  activity variable for the Unemployed (21+) category of recipients.30 The greater the proportion of fortnights a recipient in this category is listed  as being  engaged in job search under  the Activity  Test, the higher  the risk ratio. Second, repeat spells  of income support significantly reduce the risk ratio indicating that if the recipient has previously been  on income support, the probability of exiting the income support system  falls.

Table 12: Proportional hazards model, unemployed 21+, LDS 1% Sample.
Variable Males Females
Parameter
Estimate
Wald
Chi-Square
Pr> Chi-Square Risk
Ratio
Parameter
Estimate
Wald
Chi-Square
Pr> Chi-Square Risk
Ratio
Age
Age –0.0161 9.05 0.00 0.98 –0.0281 8.53 0.00 0.97
Age squared 0.0001 2.55 0.11 1.00 0.0002 3.57 0.06 1.00
Country of birth
Main English speaking country 0.1261 20.63 0.00 1.13 0.0758 2.75 0.10 1.08
Non English speaking country –0.0182 0.66 0.42 0.98 0.0086 0.06 0.80 1.01
Aboriginal and Torres Strait  Islander status
Aboriginal  and Torres Strait Islander –0.1286 6.86 0.01 0.88 –0.1002 1.20 0.27 0.91
Housing  Tenure
Purchaser 0.1795 15.13 0.00 1.20 0.1320 2.26 0.13 1.14
Other home owner 0.2204 9.01 0.00 1.25 –0.0839 0.49 0.48 0.92
Private  rent –0.0203 0.61 0.44 0.98 0.0438 1.23 0.27 1.05
Government rent –0.2914 33.37 0.00 0.75 –0.3073 15.31 0.00 0.74
Other rent –0.1036 16.98 0.00 0.90 0.0025 0.00 0.95 1.00
Rent amount
Rent ($) –0.0004 17.57 0.00 1.00 –0.0003 6.45 0.01 1.00
Rent Assistance sharer status
Sharer –0.4105 135.53 0.00 0.66 –0.3981 70.05 0.00 0.67
Dependent children
Aged 0–6 –0.0544 4.67 0.03 0.95 –0.0590 0.36 0.55 0.94
Aged 7–12 0.0352 0.73 0.39 1.04 0.0357 0.08 0.77 1.04
Aged 13–15 0.0048 0.01 0.91 1.01 –0.2332 3.40 0.07 0.79
Aged over 15 0.0913 1.02 0.31 1.10 0.0540 0.09 0.76 1.06
Income
Average  unearned income 0.0003 9.00 0.00 1.00 0.0005 9.17 0.00 1.00
Average  part-time  earnings during  the spell 0.0002 24.50 0.00 1.00 0.0000 1.41 0.24 1.00
Income support variables
Partner  receives an unemployment-related benefit –0.0651 2.50 0.11 0.94 –0.0466 1.15 0.28 0.96
Average  job search  activity 2.5012 3527.00 0.00 12.20 2.1757 1423.00 0.00 8.81
Duration on income support > 30 days –0.0012 0.00 0.95 1.00 0.0375 1.43 0.23 1.04
Repeat  spells  on income support during  sample  period –0.1361 65.49 0.00 0.87 –0.1938 5.77 0.00 0.82
N (spells) 20145     8401    
Censored spells 4398     1873    
Testingglobal null hypothesis No B B Model  chi-square No B B Model  chi-square
Likelihood ratio test 287887.6 282942.2 4945.4 with 22 DF (p=0.0001) 107958.5 105822.5 2136.0 with 22 DF (p=0.0001)
Score test     4049.5 with 22 DF (p=0.0001)     1775.0 with 22 DF (p=0.0001)
Wald test     4253.0 with 22 DF (p=0.0001)     1779.0 with 22 DF (p=0.0001)

Source: LDS 1% Sample

 

Table 13: Proportional hazards function, Parenting  Payment recipients, LDS 1% Sample.
Variable Female Parenting Single Female Parenting Partnered
Parameter
Estimate
Wald
Chi-Square
Pr> Chi-Square Risk
Ratio
Parameter
Estimate
Wald
Chi-Square
Pr> Chi-Square Risk
Ratio
Age
Age 0.0682 8.25 0.00 1.07 0.129 60.39 0.00 1.14
Age squared –0.0010 7.67 0.01 1.00 –0.002 56.27 0.00 1.00
Country of birth
Main English speaking country 0.0834 1.06 0.30 1.09 0.144 6.01 0.01 1.16
Non English speaking country –0.0311 0.16 0.69 0.97 –0.312 55.13 0.00 0.73
Aboriginal  and Torres Strait Islander  status
Aboriginal  and Torres Strait Islander 0.0448 0.16 0.69 1.05 –0.042 0.15 0.70 0.96
Housing  Tenure
Purchaser 0.0741 0.58 0.44 1.08 0.160 7.16 0.01 1.17
Other home owner –0.0547 0.17 0.68 0.95 0.213 2.59 0.11 1.24
Private  rent 0.2138 9.31 0.00 1.24 –0.043 0.68 0.41 0.96
Government rent –0.0740 0.50 0.48 0.93 –0.432 29.45 0.00 0.65
Other rent 0.0858 1.08 0.30 1.09 –0.133 3.36 0.07 0.88
Rent amount
Rent ($) –0.0009 18.75 0.00 1.00 0.000 3.66 0.06 1.00
Rent assistance sharer status
Sharer –0.7957 0.63 0.43 0.45 –1.234 1.51 0.22 0.29
Dependent children
Aged 0–6 –0.5437 39.45 0.00 0.58 –0.380 50.51 0.00 0.68
Aged 7–12 –0.5267 37.89 0.00 0.59 –0.396 50.63 0.00 0.67
Aged 13–15 –0.1034 1.93 0.16 0.90 –0.149 10.81 0.00 0.86
Aged over 15 –0.3303 3.90 0.05 0.72 –0.008 0.01 0.94 0.99
Income
Average  unearned income 0.0012 5.88 0.02 1.00 0.000 37.97 0.00 1.00
Average  part-time  earnings during  the spell 0.0012 199.87 0.00 1.00 0.001 47.01 0.00 1.00
Income  support variables
Partner  receives an unemployment-related benefit –13.539 0.00 0.99 0.00 0.151 22.01 0.00 1.16
Average  job search  activity –0.543 1.87 0.17 0.58 –1.485 21.00 0.00 0.23
Duration on income support > 30 days –0.091 0.35 0.56 0.91        
Repeat  spells  on income support during  sample  period 0.031 0.39 0.53 1.03 0.006 0.03 0.86 1.01
N (spells) 3448       5872  
Censored spells 1694       1791  
Testingglobal null hypothesis No B B Model chi-square No B B Model chi-square
Likelihood  ratio test 26490.1 26202.3 287.8  with  22 DF (p=0.0001) 65352.805 64957.5 395.3  with  21 DF (p=0.0001)
Score test     333.4  with  22 DF (p=0.0001)     410.3  with  21 DF (p=0.0001)
Wald test     329.6  with  22 DF (p=0.0001)     401.0  with  21 DF (p=0.0001)

Source: LDS 1% Sample.

6.4 Summary

The most important policy issue  we address  in this paper  is the question of whether obtaining work  while on income support improves the chances of final exit  from income support. If so, then the evidence supports the stepping-stone thesis  that part-time  employment provides a potentially important stepping-stone to a job with  sufficiently high wages that the individual can exit  the income support system. We tested  the stepping-stone hypothesis by examining movements in the time series  exit  rate and by considering the role of labour  earnings in a statistical model  of the hazard  rate.

The time series  exit  rate is given  as the proportion of income support recipients at time period t who  are not on income support at time period  t + 1. A key conclusion from our times series analysis of the transition from income support to non-support is that the exit  rate for those income support recipients who  are employed at a given  point of time is generally above the
exit  rate for non-earners. This result  applies at the aggregate level  and for both women and men but is particularly pronounced among  the Unemployed (<21) and Unemployed (21+) categories. The gap in the exit  rate between earners and non-earners for the Unemployed (<21) and Unemployed (21+) recipient groups  is around  6 to 10 per cent  in the case of the short-run exit  rate (exit  from income support within a month). This gap rises to 15 per cent  in the case of the long-run exit  rate (exit  from income support within a year). The difference in exit  rates between earners and non-earners narrows considerably, however, in the case of Parenting Single recipients. This result  is in line with  our previous hypothesis that the voluntary part-time employment rate is considerably higher  for this category of income support recipients than for the prime-age unemployed.

In hazard  rate modelling, the focus is not on the proportion of income support recipients who exit  the income support system  at a particular calendar time point but on exit  from income support at fortnightly points  within given  income support spells. The hazard  rate for a given fortnight extract date within an income support spell  is derived by dividing the number  of recipients in a chosen income support category who  were observed to exit  from income support in that fortnight by the number  of recipients who  potentially could  have exited in that fortnight. Our estimated models  of the hazard  rate confirm  the important role for exits  of employment episodes, while on income support, for the unemployed. Part-time earnings have a significant positive effect on the hazard  rate controlling for all other  effects. Our results, therefore, strongly support the stepping-stone hypothesis. The take-up of employment options among  income support recipients aids the movement off income support and into longer-hours employment for those  recipients who  prefer  such jobs.

[ Return to Top   Return to Section ]

7 Summary and policy implications

This paper  has utilised the FaCS LDS 1% Sample  to examine how  income support recipients interact with  the Australian  labour  market. The LDS 1% Sample  is a unique data set because of it contains detailed longitudinal information on income support recipients over an extended period of time.

The purpose of this concluding chapter is to present a summary statement of the results  and outline  the policy implications of our analysis. Our summary statement links results  from various  chapters of the report  to provide  a typology of the way  recipients from the unemployed and families  with  children income support programs engage with  the labour  market.

The take-up of employment positions while on income support

Across all categories of income support recipients studied in this paper, take-up of employment positions, while on income support, is highest among  female  Parenting Single recipients.
The proportion of female  Parenting Single recipients receiving earned income (the employment rate)  was around  26 per cent  in 1995  but had risen  to 29 per cent  in June 1999. The corresponding employment rate for female  Unemployed (21+) recipients, was not only lower but had also fallen over the period  covered by the LDS 1% Sample; from 24 per cent  in 1995  to around  to 22 per cent  at the end of the period. A similar  decline from a lower initial base is evident among  male Unemployed (21+) recipients. Among Unemployed (<21) recipients, the decline in the proportion of recipients in paid employment was even  more pronounced over the 1995  to 1999  period. Employment rates among  older  unemployed recipients are very low and an obvious  concern for policy makers.

Why might  employment rates among  Parenting Single income support recipients be higher than for other  groups?  There are a number  of possible explanations, but two important causes relate  to the structure of applicable income tests and differences in labour  supply preferences between Parenting Single  recipients and recipients in other  categories. In terms  of the former effect, entitlement to income support is maintained for a greater range  of earnings and at higher levels  for Parenting Single  recipients as compared to allowees. In terms  of the role of preferences, the labour  supply preference structure for Parenting Single  recipients may be more skewed towards increased hours devoted to non-work  activities (such  as child  care responsibilities). This means  that Parenting Single  recipients are more likely  to optimally locate themselves at labour  supply positions consistent with  the maintenance of their  income support payment. Voluntary  part-time  employment, while on benefit, is more likely  to characterise the Parenting Single group  than the unemployed. We would, therefore, expect higher  rates of take- up of employment positions together with  a more stable  employment profile  while on income support among  Parenting Single  recipients as compared to other  groups  in this study.

As noted, employment rates among  income support recipients, other  than those  in the Parenting Single category, fell over the 1995  to 1999  period. Winding back Working Nation initiatives in 1997  and 1998  may have had an effect on the availability of subsidized part-time employment positions for unemployed income support recipients. Mention  should  also be made  of the impact  of changes to the structure of income support arrangements over the LDS 1% Sample  period  (for example, the Earnings Credit  Scheme was abolished in 1997), and the cumulative effect of non-indexation of withdrawal brackets for non-pension income support payments. Both of these  factors  would  have contributed to a decline in employment rates for the relevant groups.

It is, however, possible to view  the decline in employment rates among  the unemployed in a positive light, given  that the unemployment rate fell in the period  from the beginning of 1998. The improving state of the labour  market  may have resulted in a greater take-up of longer- hours jobs among  the unemployed rather  than shorter-hours jobs as may have been  the
case for Parenting Single  recipients (given  a presumed higher  rate of voluntary part-time employment). Unemployed income support recipients who  may otherwise have accepted short-hours  jobs may have moved  off income support altogether and into positions with longer  hours and higher  pay. As such the employment rate, which refers  to employment while on income support, falls while the exit  rate rises.

Our modelling analysis of the take-up of employment positions while on income support indicates that Parenting Single  recipients have a higher  probability of being  employed while on income support than do other  recipients having  controlled for other  determinants of earned income receipt. These determinants include the number  of dependent children in the case of female  recipients (negative effect), age (positive, but diminishing effect), Aboriginal  and Torres Strait Islander  status (negative), non-English speaking background (negative), government rental status (negative), and job search  activity under  the Activity  Test (very  strongly positive).

Our analysis also revealed that the earnings distribution function  for Parenting Single income support recipients lies well  to those  right of those  for other  income support payment groups; a natural  outcome given  the greater earnings range  available under  the pensions income test relative to the allowance income test. Research studies  have consistently pointed to the higher
incidence of poverty among  sole parents in the Australian  community. Relatively high take-up of part-time  employment positions and a strong  earnings profile  helps  to significantly ameliorate both the incidence and severity of poverty among  this group. One further  result  that is of interest is the gender earnings gap in favour of women in the LDS 1% Sample. This earnings gap may reflect  an hours effect rather  than a wage  rate effect (we  have no information on employment hours in the data) but nevertheless the result  is of some interest in light of the
well-recorded earnings disadvantage felt in the labour  market  generally by women.

Income support spell and employment episode histories

The distinction between the labour  market  experience of unemployed income support recipients and Parenting Single  recipients is put into sharper relief  when  we examine their income support spell  and employment episode history  profiles. The Unemployed (<21) and Unemployed (21+) groups, typically have both shorter  income support spells  and shorter employment episodes than can be found for each  of the remaining income support categories. Moreover, as the income support spell  lengthens, the gap between the mean  duration of employment episodes for Parenting Single  recipients and those  for the Unemployed (<21) and Unemployed (21+) groups  increases. The female  Parenting Single recipient, who  gets a part- time job while on income support, spends, on average, twice as much  time in employment as male Unemployed (21+) recipients (who  obtain  employment while on income support).

Likewise, there  is a tendency for the Unemployed (<21) and Unemployed (21+) groups  to experience a somewhat higher  degree of job churning (multiple employment episodes) than can be found among  Parenting Single recipients. In short, the Parenting Single  pattern of engagement with  the labour  market  is much  more one of long-duration spells  of income support and stable  patterns of income receipt than is evident for the two main unemployed income support payment recipient groups. This evidence is consistent with  our original hypothesis of higher  rates of voluntary part-time  employment among  Parenting Single income support recipients.

One point of concern to policy-makers from our results  is the relatively low take-up of part-time employment positions among  the unemployed, in the medium to long-duration spells  of income support categories. To take one example, among  male Unemployed (21+) recipients with income support spells  lasting  more than two years, 46 per cent  had not experienced a single employment episode during  their  spell  on income support. This is a point of significant concern as the longer  the unemployed person  remains away  from any sort of employment, the more their  skills and mental  health  can deteriorate. A focus of any employment-related strategy for the long-term  unemployed people in such a position is the provision of wage  subsidies to enable these  people to engage in the formal labour  market.

Our results  also show  that Unemployed (<21) and Unemployed (21+) recipients may typically experience two or three  income support spells  over a five year  period. They also typically do not experience an employment episode in these  multiple spells  on income support.

Earnings and exit from income support

A key conclusion from our analysis of the transition from income support to non-support is that the exit  rate for those  who  are employed at a given  point of time is generally above the exit  rate for non-earners. This result  applies at the aggregate level  and for both women and men but is particularly pronounced among  the Unemployed (<21) and Unemployed (21+) recipient groups. The gap in exit  rates between earners and non-earners for the Unemployed (<21) and Unemployed (21+) recipient groups  is around  6 to 10 per cent  in the case of the short-run exit rate (exit  from income support within a month). This gap rises to 15 per cent  in the case of the long-run exit  rate (exit  from income support within a year). The gap between earners and non- earners narrows considerably, however, in the case of Parenting Single recipients.

Our estimated models  of the hazard  rate confirm  the important role of employment episodes while on income support for the unemployed. Part-time earnings have a significant positive effect on the hazard  rate controlling for all other  effects. Our results  strongly support the stepping-stone analogy. Part-time employment options  will  aid in the movement off income support and into longer-hours employment for those  income support recipients who  prefer such jobs.

[ Return to Top   Return to Section ]

Endnotes

  1. The term ‘earned income’ is used throughout this paper  to refer to income derived from employment in the labour  market. This definition is consistent with  that used in the FaCS LDS 1% Sample  dataset.
  2. While  the removal  of a 100 per cent  benefit  withdrawal rate reduces poverty trap problems for existing income support recipients, trade-offs do exist. With the introduction of a moderate taper, individuals who  previously may have been  ineligible for an income support payment, because earnings were marginally above the level  at which entitlement is lost, now  become eligible for support. This results  in labour  supply disincentive effects  for those previously on the margin  of income support. Such income support recipients are drawn into the income support payment system  and work  fewer  hours than they  did previously.
  3. Hours-constrained employees are said to be involuntarily ‘part-time’ employed workers.
  4. The LDS 1% Sample  makes  a distinction between ‘earned income’ and ‘unearned income’. Earned income is defined  as income from salary  and wages. Unearned income is income received by an income support payment recipient from sources other  than salary  and wages. Our analysis is concerned solely  with  the earned income component.
  5. The paper  covers  the period  1995  to 1999. Reforms to the income support system  since that time are not addressed in this paper.
  6. Creedy  & Scutella  (2000) examine, in considerable detail, the way  in which the structure of income tests and the tax system  impact  on the distribution of earnings. Their paper  is also based  on the LDS 1% Sample. The reader is referred to this source for a fuller treatment of the topic.
  7. At times we use the term ‘part-time employment’ loosely  in this report  to refer to the take- up of short-hour  jobs that would  not result  in the loss of an income support payment as a result  of the application of income tests.
  8. Junor (1998) suggests that casual  part-time  employment contributed 55.9  per cent  of the increase in the total number  of jobs held by men in the period  1987  to 1997; permanent part-time  employment contributed a further  7.3 per cent  to the change in total jobs. For women, the respective figures  are 38.9  per cent  and 30.9  per cent.
  9. The figures  quoted  all refer to April 1999  (see  Australian  Bureau  of Statistics, Labour Force, Australia, April1999, Cat. No. 6203.0.
  10. See Chapman, Jordan, Oliver, & Quiggan  (2000) for an interesting account of the way  in which a dynamic, rather  than static, choice-theoretic framework may affect labour  supply decisions among  income support recipients.
  11. In more formal language, there  exists  a price  substitution effect in favour of increased supply of labour  hours when  the withdrawal rate falls. There is, however, no certainty that optimal  labour  supply will  increase with  a reduction in the benefit  withdrawal rate. Whether this happens depends on both the price  substitution effect and the so-called income effect. As the net relative wage  rate rises for income support recipients from a relaxation of the income test, so too does income. Higher incomes may induce a preference for non-work  activities (particularly for those  with  young  dependent children). Labour supply increases provided that the substitution effect outweighs the income effect (if the latter  is negative in terms  of increased work  hours).
  12. Beyond  the efficiency and labour  market  effects  noted  above, the relaxation of the poverty trap constraint ensures higher  levels  of gross income for those  engaged in part-time employment. This acts to support the poverty alleviation and reduction in income inequality objectives of social  policy programs.
  13. One exception should  be noted. In the case of the Mature Age Allowance, paid to unemployed persons 60 and over, an allowance first paid prior to July 1996  is assessed against  the pension income test.
  14. The date for the METR simulation of end-1996  is chosen for convenience as an analysis of METRs for this point in time was conducted as part of study  by Flatau and Wood (2000) on the treatment of housing wealth in the measurement of comprehensive income.
  15. The METR does not equal  84 per cent  at this level  (or a 50 per cent  withdrawal rate plus a 34 per cent  marginal tax rate)  because rebates applicable to the Sole Parent Pension recipient have the effect of reducing the net tax paid and hence the value  of the METR.
  16. See Australian  Bureau  of Statistics 1996, Employee Earnings and Hours, Australia, May 1996, Cat. No. 6306.0.
  17. On an after-housing cost basis, receipt of rental  assistance would  (assuming relatively low rents)  push income units marginally above the corresponding after-housing cost poverty line.
  18. Whether or not the amounts  recorded for a given  income support payment recipient match the actual  earned income stream  of the individual over time depends on a number  of factors. These factors  include: the timeliness and accuracy of the fortnightly information supplied to Centrelink by the recipient; the efficient processing of that information by Centrelink; administrative practices regarding the recording of such information by Centrelink; and the maintenance of that information in data files over time.
  19. See the Department of Social Security 1997–98 Annual Report for evidence that supports the claim  that the removal  of the Earnings Credit  Scheme reduced the incentive for income support recipients to engage in the labour  market
  20. As noted  previously, income support recipients who  are no longer  receiving an income support payment may be retained in the LDS 1% Sample  for a period. This practice appears to be more prevalent in the later years  of the LDS 1% Sample  than the earlier years, perhaps reflecting changes in administrative practices over time. The estimates of earnings contained in figures  15 to 20 are based  on income support recipients who  are receiving a positive income support payment. One major reason  for non-receipt of benefit  is, of course, too high a level  of private income. It comes  as no surprise, therefore, that mean  and median earnings are considerably higher for the non-benefit category of income support recipients than for those  receiving an income support payment.
  21. In the case of unemployed Newstart allowees who  had partners receiving Parenting Payment, entitlement to the latter  income support payment would  not be lost until the recipient was earning $816  per fortnight.
  22. Note the sample  selection effect here. We only have information on the earnings of current income support recipients. We do not have information on the earnings of former income support recipients.
  23. As explained by Creedy  & Scutella  (2000), a wide  variety  of labour  supply responses are in fact possible in response to kinks in the budget  line facing  the income support recipient. Our presentation at this point is, therefore, fairly simplistic. Moreover, as explained in Chapter 2, what  matters  to the income support recipient is not simply  the withdrawal rate structure they  face but the combination of both income tests and tax effects. We summarize these  joint effects  in the METR structure. The important point is that income support recipients are facing  both tax effects  and income test effects  at relatively low earnings levels. This means  that it is difficult  to distinguish between the two effects.
  24. The interested reader is referred to the work  on income support spell  churning by Dawkins, Harris & Loundes (2000). This paper  is also based  on the LDS 1% Sample.
  25. We must continue to emphasise the point that our estimates of employment episode duration refer only to periods of employment while on income support. An unemployed person  may have a relatively short employment episode recorded when  in point of fact that recipient moved  from a part-time  job while on income support to longer  hours in the same job (leading to exit  from the income support system).
  26. To extend our transition analysis, we estimated logistic regression models  of transition from income support to a non-support position by the end of one, two and three  year  period points. The dependent variable Y takes  the value  1 if an individual on income support in July1996 was no longer  on income support one year, two years  or three  years  later. A value of zero is taken  if the individual did not exit  income support. For both women and men, the earnings variable is positive and significant, indicating that as labour  income rises so too does the probability of exiting income support. A more detailed analysis of exit  behaviour using  the more conventional hazard  analysis techniques is presented in Chapter 6.
  27. Note that the LDS 1% Sample  appears to be in error in terms  of earnings for male income support recipients in the Unemployed (<21) category for the first fortnight in June 1996. This outcome affects the exit  rate for this extract date.
  28. We end with  the 100th  fortnight in the income support spell  as the number  of income support recipients remaining in the system  at that point is relatively small, leading to a highly  erratic function  at that point. In any case there  are only 104 potential fortnights in the LDS 1% Sample  given  that the sample  period  is from June 1995  to July 1999  (having deleted all existing income support spells  at the start of the LDS 1% Sample  in 1995).
  29. A single  risk hazards  model  has been  estimated with  a time-constant earnings variable. In McCall’s  (1996), model  of part-time  employment and social  insurance arrangements, a spell of unemployment may end either in part-time  employment or in full-time employment (that is, competing risks rather  than single  risks are identified). The re-employment hazard in this case is given  by:
    (t)=   f(t) +  p(t).
    where  f(t) and   p(t) are the full-time and part-time  re-employment hazards  respectively. The model  is estimated using  time-varying coefficients. Justification for the use of a model with  time varying  coefficients is provided from theory  and from McCall’s  findings  that the disregard had the greatest effect at the beginning of the unemployment spell. The null hypothesis of time-constant coefficients was rejected by McCall at the 1 per cent significance level.
  30. The job search  activity variable is derived from the Activity  Test variable in the LDS 1% Sample  and is relevant only for those  recipients in the Newstart Allowance,Youth Training/Youth  Allowance and Austudy  payment categories.

[ Return to Top   Return to Section ]

References

Atkinson, A.B. & Micklewright, J. 1991,‘Unemployment Compensation and Labor Market Transitions: A Critical  Review’, Journal of Economic Literature, vol. 24, pp. 1679–1727.

Barrett, G.F. & Cragg, M.I. 1998,‘An Untold Story: The Characteristics of Welfare  Use in British Columbia’, Canadian Journalof Economics, vol. 31, pp. 165–188.

Blanchflower, D. & Corry, B. 1986,‘Part-time Employment in Great Britain: An Analysis Using Establishment Data’, Department of Employment Research Paper, no. 57.

Blanks, R. 1990,‘Understanding Part-time Work’, Research in Labor Economics, vol. 11, pp. 137–158.

Bradbury, B. 1999,‘Tax Theory  and Targetting: A Survey’, Social Policy Research Centre Discussion PaperSeries, no. 100.

Chapman, B., Jordan, J., Oliver, K. & Quiggan, J. 2000,‘The Unemployment Trap Meets the Age- Earnings Profile’, mimeo.

Creedy, J. & Scutella, R. 2000,‘Means-tested Benefits, Incentives and Earnings Distributions’, paper  presented to the Department of Family and Community Services Workshop on the LDS 1% Sample, Canberra, ACT.

Dawkins, P., Harris, M.N. & Loundes, J. 2000,‘Repeated Spells on Benefits: An Analysis of “Churning”, using  the FaCS Longitudinal Administrative Data Set’, paper  presented to the Department of Family and Community Services Workshop on the LDS 1% Sample, Canberra, ACT.

Dawkins, P. & Norris, K. 1995,‘The Growth of Part-time Employment in Australia  and the United Kingdom, 1978–93’, Labour Economics and Productivity, vol. 7, pp. 2–27.

Flatau, P., Galea, J. &Petridis, R. 2000,‘Mental Health and Wellbeing and Unemployment’, The Australian Economic Review, vol. 33, pp. 161–181.

Flatau, P. & Simpson, M. 1996,‘Part-time Youth Employment and Training: Evidence  from the Australian Youth Survey’, Labour Economics and Productivity, vol. 8, pp. 131–162.

Flatau, P. & Wood, G.A. 2000,‘Comprehensive Income  Measures, Housing Equity, and Tax- Transfer Effects’, Australian Economic Papers, vol. 39, pp. 325–344.

Gregory, R. & Duncan, R. 1980,‘High Teenage Unemployment: The Role of Atypical  Labour Supply  Behaviour’, Economic Record, vol. 56, pp. 316–330.

Gritz, R.M. & MaCurdy, T. 1997,‘Measuring the Influence of Unemployment Insurance on Unemployment Experiences’, Journalof Business and Economic Statistics, vol. 15, pp. 130–152.

Haveman, R. 1996,‘Reducing Poverty While  Increasing Employment: A Primer  on Alternative Strategies, and a Blueprint’, OECD Economic Studies, no. 26, pp. 7–42.

Junor, A. 1998,‘Permanent Part-time Work: New Family-friendly Standards  or High Intensity Cheap  Skills?’, Labour and Industry, vol. 8, pp. 77–95.

Kalb, G. 1998,‘An Australian  Model for Labour Supply  and Welfare  Participation in Two-adult
Households’, Social PolicyResearch Centre, Discussion  Paper Series, no. 82. Kaspura, A.J. (ed.)  1984, Labour Force Participation in Australia, AGPS, Canberra.
Katz, L.F. & Meyer, B.D. 1990,‘The Impact  of the Potential Duration of Unemployment Benefits on the Duration of Unemployment’, Journal of Public Economics, vol. 41, pp. 45–72.

Keifer, N.M. 1988,‘Economic Duration Data and Hazard Functions’, Journal of Economic Literature, vol. 26, pp. 646–679.

Keifer, N.M. & Neuman, G.R. 1989, Search Models and Applied Labor Economics, Cambridge University Press.

Killingsworth, M.R. 1983, Labour Supply, Cambridge University Press. Lancaster, T. 1990, The Econometric Analysis of Transition Data, Cambridge University Press. Leeves, G.D. 1997,‘Labour Market Gross Flows and Transition  Rates’, The Economic and Labour Relations Review, vol. 8, pp.110–127.

Lewis, H. 1990, Part-time work: trends and issues, AGPS, Canberra.

McCall, B.P. 1995,‘The Impact  of Unemployment Insurance Benefit Levels on Recipiency’, Journal of Business and Economic Statistics, vol. 13, pp. 189–198.

McCall, B.P. 1996,‘Unemployment Insurance Rules, Joblessness, and Part-time Work’, Econometrica, vol. 64, pp. 647–682.

McCall, P.B. 1997,‘The Determinants of Full-time versus  Part-time Reemployment following Job Displacement’, Journal of Labor Economics, vol. 15, pp. 714–733.

Meyer, B.D. 1990,‘Unemployment Insurance and Unemployment Spells’, Econometrica, vol. 58, pp. 757–782.

Narendranathan, W., Nickell, S. & Stern, J. 1985,‘Unemployment Benefits  Revisited’, The Economic Journal, vol. 95, pp. 307–329.

Narendranathan, W. & Stewart, M.B. 1993,‘How Does the Benefit Effect Vary as Unemployment Spells Lengthen?’, Journal of Applied Econometrics, vol. 8, pp. 361–381.

National Board of Employment, Education  and Training  NBEET 1992, Disadvantaged Jobseekers: Casual, Part-time and TemporaryWork, Commissioned Report  no.18, Canberra. OECD 1994, The OECD Jobs Study: Evidence and Explanations, OECD, Paris.
OECD 1997, The OECD Jobs Strategy: Making Work Pay, Taxation, Benefits, Employment and Unemployment, OECD, Paris.

Robertson, P. 1989,‘Some Explanations For The Growth of Part-time Unemployment [sic] In Australia’, Australian Bulletin of Labour, vol. 15, pp. 384–399.

Romeyn, J. 1992, Flexible Working Time: Part-time and Casual Employment, Department of Industrial Relations, Canberra.

Rubery, J., Earnshaw, J. & Burchell, B. 1991, New Forms and Patterns of Employment: The Role of Self  Employment in Britain, Report  for the Zentrum  for Europaisches Rechtspolitik, Bremen.

Sadler, C. & Aungles, P. 1990,‘Part-time employment growth in Australia, 1978–1989’, Australian Bulletin of Labour, vol. 16, pp. 286–296.

Stevens, A.H. 1997,‘Persistent Effects of Job Displacement: The Importance of Multiple  Job Losses’, Journal of Labor Economics, vol. 15, pp. 165–188.

Trivedi, P.K. & Baker, G.M. 1983,‘Unemployment in Australia: Duration and Recurrent Spells’, The Economic Record, June, pp. 132–148.

Whiteford, P., Bradbury, B. & Saunders, P. 1989,‘Poverty Traps in the Australian  Social Security System’, Economic Analysis and Policy, vol. 19, pp. 1–27.

Whynes, D. 1993,‘The Poverty  Trap’, in Barr and Whynes (eds)  Current Issues in the Economics of Welfare, MacMillan.

Wooden, M. 1996,‘The Youth Labour Market: Characteristics and Trends’, Australian Bulletin of Labour, vol. 22, pp. 137–160.

[ Return to Top   Return to Section ]

List of Figures

1     The effect of a 100 per cent  withdrawal rate on hours of employment

2     The effect of an earnings disregard and a less than 100 per cent  withdrawal rate on hours of employment

3     Gross labour  earnings and income unit marginal effective tax rates (METR), Parenting Single plus two kids, end 1996

4     Gross labour  earnings and income unit marginal effective tax rates (METR), Unemployed(21+), end 1996

5     Gross earnings and income unit marginal effective tax rates (METR), Unemployed (21+) earner plus spouse  receiving Parenting Payment, two kids, end 1996

6     The ratio of disposable income to the before-housing Henderson Poverty  Line, end 1996

7     Unemployed income support payment recipients, men, LDS 1% Sample

8     Unemployed income support payment recipients, women, LDS 1% Sample

9     Parenting income support payment recipients, men, LDS 1% Sample

10   Parenting income support payment recipients, women, LDS 1% Sample

11   Employment rates, unemployed income support payment recipients, women, LDS 1% Sample

12   Employment rates, unemployed income support payment recipients, men, LDS 1% Sample

13   Employment rates, parenting income support payment recipients, women, LDS 1% Sample.

14   Employment rates, parenting income support payment recipients, men, LDS 1% Sample

15   The distribution of earnings, Unemployed (<21), LDS 1% Sample

16   The distribution of earnings, Unemployed (21+), LDS 1% Sample

17   The distribution of earnings, Unemployed (60+) LDS 1% Sample

18   The distribution of earnings, Partner Allowance, LDS 1% Sample

19   The distribution of earnings, Parenting Single, LDS 1% Sample

20   The distribution of earnings, Parenting Partnered, LDS 1% Sample

21   Distribution of income support spells  by payment category, women, LDS 1% Sample

22   Distribution of income support spells  by payment category, men, LDS 1% Sample

23   Employment episodes by spells  of income support, 1995  cohort, Unemployed (<21), women, 1995  to 1999  LDS 1% Sample

24   Employment episodes by spells  of income support, 1995  cohort, Unemployed(<21), men, 1995  to 1999, LDS 1% Sample

25   Employment episodes by spells  of income support, 1995  cohort, Unemployed (21+), women, 1995  to 1999, LDS 1% Sample

26   Employment episodes by spells  of income support, 1995  cohort, Unemployed(21+), men, 1995  to 1999, LDS 1% Sample                                                      

27   Employment episodes by spells  of income support, 1995  cohort, Unemployed (60+), men, 1995  to 1999, LDS 1% Sample

28   Employment episodes by spells  of income support, 1995  cohort, Partner Allowance, women, 1995  to 1999, LDS 1% Sample

29   Employment episodes by spells  of income support, 1995  cohort, Parenting Single, women, 1995  to 1999, LDS 1% Sample

30   Employment episodes by spells  of income support, 1995  cohort, Parenting Single, men 1995  to 1999, LDS 1% Sample

31   Employment episodes by spells  of income support, 1995  cohort, Parenting Partnered, women, 1995  to 1999, LDS 1% Sample                                             

32   Employment episodes by spells  of income support, 1995  cohort, Parenting Partnered, men, 1995  to 1999, LDS 1% Sample

33   Short-run exit  rates for Unemployed (<21) income support payment recipients, moving  average estimates, LDS 1% Sample

34   Long-run exit  rates for Unemployed (<21) income support payment recipients, moving  average estimates, LDS 1% Sample

35   Short-run exit  rates for Unemployed (21+) income support payment recipients, moving  average estimates, LDS 1% Sample

36   Long-run exit  rates for Unemployed (21+) income support payment recipients, moving  average estimates, LDS 1% Sample

37   Short-run exit  rates for Unemployed (60+) income support payment recipients, moving  average estimates, LDS 1% Sample

38   Long-run exit  rates for Unemployed (60+) income support payment recipients, moving  average estimates, LDS 1% Sample

39   Short-run exit  rates for Partner Allowance income support payment recipients, moving  average estimates, LDS 1% Sample

40   Long-run exit  rates for Partner Allowance income support payment recipients, moving  average estimates, LDS 1% Sample

41   Short-run exit  rates for Parenting Single income support payment recipients, moving  average estimates, LDS 1% Sample

42   Long-run exit  rates for Parenting Single income support payment recipients, moving  average estimates, LDS 1% Sample

43   Short-run exit  rates for Parenting Partnered income support payment recipients, moving  average estimates, LDS 1% Sample

44   Long-run exit  rates for Parenting Partnered income support payment recipients, moving  average estimates, LDS 1% Sample

45   Hazard function  for Unemployed (21+) income support payment recipients, LDS 1% Sample

46   Hazard function  for parenting income support payment recipients, LDS 1% Sample

[ Return to Top   Return to Section ]

Last updated: