The Transmission of Education across Generations: Evidence from Australia

1 Introduction

Education enhances human development, provides individuals with opportunities to improve their earnings and promotes the general welfare through the positive externalities associated with it. Studies on the relationship between income and education show that there is a considerable and significant relationship between the two.^[1] Moreover education is seen as the “ultimate equalizer” of economic opportunity and income, leading to increased social mobility and economic progress (Bowles 1972; Becker and Tomes 1979; Black, Devereux, and Salvanes 2005b; Reeves 2014). From a policy perspective the nature and extent of the persistence of education attainment across generations have important implications on the functioning and role of a country’s education system in making the quality of and access to education more equitable (Chechi, Ichino, and Rustichini 1999).

Australia provides an interesting case in the study of intergenerational education mobility. Like most other industrial countries, education participation has increased significantly across all levels over the last five decades. The proportion attaining a post-secondary education increased steadily for both men and women, with women overtaking men in the late 1970s (Figure 1). Currently secondary education is primarily government funded which effectively allows free access to schooling at this level. However, around 35% of secondary school students attend fee-paying private schools that are partially funded by the government (ABS 2014).^[2] This prominence of private secondary schools sets Australia apart from most Organisation for Economic Co-operation and Development (OECD) countries where government funded public schooling is the dominant sector.^[3] Recent studies have examined the increasing social stratification within the education system and the resulting inequities in Australia (Smyth 2014 and Perry and Lubienski 2014). At the tertiary level, almost all the universities are public^[4] where fees are set and capped by the government and supported by an income-contingent student loan scheme. Proposals stemming from the 2014 Australian federal budget will effectively replace this with a fully deregulated, demand-driven higher education sector while reducing government funding at all levels of education. While the long term outcomes of such significant policy changes are yet to be seen, studies have speculated that increasing costs will disproportionately burden low income families (NATSEM 2014; Whiteford and Nethery 2014).

Figure 1:

Proportion with a post-secondary qualification by year of birth.

The Australian education system has gone through a range of policy changes over the last few decades. Prior to 1964 there was no direct federal funding for secondary schools. The period from the mid-1960s to mid-1970s saw an expansion in secondary education fuelled by increased public funding along with the abolition of entrance examinations and fees in government schools which led to increasing enrolment and retention in secondary schools (ABS 2001b). This era also saw an increase in public funding to private secondary schools, which helped them to be viable alternatives to public schools. Private schools continue to receive substantial federal and state government subsidies which have primarily been used to improve school quality (lower student–teacher ratio) rather than to lower their fees (Ryan and Watson 2010).^[5] This has resulted in shifting children from high socioeconomic backgrounds to private schools, leaving public schools with a disproportionate number of children from lower socioeconomic backgrounds. Public expenditure on education fell significantly during a period of stagflation between the late 1970s and the 1980s, and a “needs-based” funding model was adopted which has been in place since then with varying degrees of changes to the funding formulae.^[6] In a major review of the Australian education system, Gonski et al. (2011) proposed 41 recommendations including substantial increase in government funding and improved equity and efficiency in the system, some of which were implemented by the then Labor government in 2013.

At the tertiary level, university fees were abolished for a brief period between 1974 and 1988, which made university education accessible to an increasing number of students (ABS 2001b). In 1988 university fees were reintroduced in the form of income-contingent student loans (Higher Education Contribution Scheme, HECS), which remain to the present day. At the time of its introduction, primary concerns were around the potential adverse effects on disadvantaged groups. However, a growing body of research suggests that the introduction of HECS in 1989 or the subsequent changes had no significant impact on participation in higher education, as the relatively disadvantaged were less likely to attend university even in the absence of student fees (Chapman 1997; Chapman and Ryan 2005; Cardak and Ryan 2006). It has been argued that the existence of a dual private-public schooling system limits the effects of any egalitarian financing in higher education, as academic performance in high school and the scores on which university entrance is based are linked to parental background (Marginson 1997; Cardak and Ryan 2006). The education system is again the focus of the current administration with major policy changes proposed at reducing government financing of education across all levels.

In the present paper, we analyze the intergenerational education mobility in Australia for those born between 1942 and 1991 using data from the 2011 Household Income and Labor Dynamics, Australia (HILDA) survey, an extensive longitudinal dataset containing information on educational attainment of individuals and their parents. The period of analysis captures the evolution of the Australian education system and the policies that shaped it. Therefore understanding changes in intergenerational education mobility over time will inform how past educational reforms and policies performed in improving equity across different socioeconomic groups. We provide measures of intergenerational transmission of education (regression coefficients and correlations) that are directly comparable with Hertz et al. (2007), and position Australia within their ranking of OECD countries,^[7] and examine regional variations within Australia at the state level. We also estimate education transition matrices (Chechi, Ichino, and Rustichini 1999; Bauer and Riphahn 2007; Chevalier, Denny, and McMahon 2009; Dauli, Demoussis, and Giannakopoulos 2010; Azam and Bhatt 2012) and a range of education mobility indicators including the Prais–Shorrocks index (Chechi, Ichino, and Rustichini 1999; Dauli, Demoussis, and Giannakopoulos 2010), upward–downward mobility indicators (Heineck and Riphahn 2009; Gandelman and Robano 2012; Chusseau and Hellier 2012) and indicators of relative education opportunities (Heineck and Riphahn 2009; Bauer and Riphahn 2007) that will enable the analysis of changes in parent–child transmission of education over time in Australia. To our knowledge, there are no comparable studies for Australia where intergenerational transmission of education is analyzed in the same way. Our results show that Australia is one of the most mobile countries in the OECD based on the measures of regression coefficients and correlations. There are however significant regional and gender variations in intergenerational education mobility over time. Daughters’ education attainment is still relatively highly correlated with their parents, compared to sons. The extent of absolute upward mobility was modest while immobility and downward mobility have remained relatively steady during the last five decades. During this period, relative education opportunities have increased over time at lower (secondary) education levels, while the trend has been comparatively stable at higher (post-secondary) levels.

The rest of the paper is organized as follows. Section 2 provides a review of existing literature. The data and theoretical aspects of intergenerational educational mobility are discussed in Sections 3 and 4, respectively, followed by a discussion of results and trends in intergenerational educational mobility in Section 5. The final section offers some concluding remarks.

2 Transmission of Education across Generations: An Overview of Existing Literature

There is a rich body of literature on the transmission of income and education across generations and the impact of education on earnings (see, for example, Becker and Tomes 1986; Blanden, Gregg, and Machin 2003; and Berhman, Pollak, and Taubman 1989). The standard specification of intergenerational schooling mobility involves a first-order Markov process where the schooling of children is a function of the schooling of parents. In this approach the coefficient of parents’ schooling represents the degree of immobility across the generations and has the advantage of being easily extended to analyze different aspects of intergenerational mobility such as differences in gender, region and socioeconomic status.^[8]

Empirical evidence on the changes in intergenerational transmission of education is, however, limited. In a comprehensive survey of the recent developments in intergenerational mobility, Black and Devereux (2011) trace the evolution of the literature and update the earlier work of Solon (1999 and 2002). As with many other studies, their survey is primarily limited to Europe and North America, reflecting the concentration of studies on these countries. Hertz et al. (2007) expand the analysis to 42 countries and present comparison of trends in intergenerational transmission of education over time. They find that the transmission of education attainment from one generation to the next, as measured by the regression coefficient, has fallen over the past 50 years, while the correlation coefficient has not changed by much. They also find considerable regional differences, with the Nordic countries displaying the lowest persistence and the Latin American countries having the highest. Daude (2011) follows a similar approach to an expanded list of 18 Latin American countries and confirms that there is low educational mobility in the region. Azam and Bhatt (2012) replicate Hertz et al. (2007) for data from India and find that while intergenerational transfer of educational attainment between fathers and sons has fallen during the last 45 years, there are significant regional variations. For Germany, Heineck and Riphahn (2009) examine the heterogeneity in intergenerational education mobility over time and find that the overall level of education and the extent of intergenerational mobility increased over time, as indicated by the indicators of upward and downward mobility. However, based on the ratio of relative educational opportunities, they also find that the probabilities of reaching high educational attainment for children from low educated parents compared to more educated parents had not changed much over the last decade. In a recent study of Italy Chechi, Fiorio, and Leonardi (2013) show that intergenerational persistence of education has declined slowly over time. Using a decomposition of the standard correlation coefficient, they find that the intergenerational persistence of education is mostly due to the fact that higher education levels are more likely to be attained by children with highly educated fathers.

Australian studies on intergenerational mobility have primarily focused on occupational status within a sociological context.^[9] Another strand of research focuses on the role of socioeconomic and family background on education attainment (Marks and McMillan 2003; Miller, Mulvey, and Martin 2005; Johnston et al. 2014; Cobb-Clark and Nguyen 2010). One of the few cross-country studies of intergenerational education mobility that includes Australia is an analysis of OECD countries with a focus on post-secondary education (de Broucker and Underwood 1998). Using data from the International Adult Literacy Survey (1994 and 1996), they find that the intergenerational correlation coefficient in Australia to be relatively low (0.29) compared to a high of 0.49 in Ireland. The current study will update these estimates for Australia with new data for 2011 and provide an in-depth analysis of the changes in long-term patterns of intergenerational transmission of education using a range of mobility indices. This study will therefore complement the growing literature on international comparisons of changes in intergenerational education mobility and fill the hitherto overlooked aspect of changes over time in Australia.

3 The Data and Descriptive Statistics

We use data from the HILDA survey for the year 2011 – a rich longitudinal survey which has been conducted annually since 2001. The survey consists of three separate components: a household questionnaire which collects household level information; a person questionnaire administered to every member of the household aged 15 years and older; and a self-completion questionnaire which includes questions on attitudes and lifestyle. The person-level questionnaire includes detailed information relating to education such as level of education, type of school (public or private), post-school qualifications, training, employment and job search activities among others. An interesting feature (and an important one for the purpose of this study) of this survey is that, in the person questionnaire, each respondent is asked to provide detailed information about their parents as part of their family history. This includes information on parental education such as their highest level of education, type of qualification, type of educational institution as well as employment details (such as occupation). We utilize this information, in addition to the respondent’s own education level, to create a dataset with comparable education levels for parents and children. This approach enables us to create a sample of parents and children which includes non-resident parents, and avoids issues of biases that arise from co-residing parent–children samples.^[10]

4 Measurement of Intergeneration Education Mobility: Conceptual Framework

A primary goal of this study is to arrive at measures of association between the education level of children and their parents and examine their evolution over time. Theories of intergenerational mobility model the parental decision to invest in their children as a combination of credit constraints and preferences (Becker 1974; Becker and Tomes 1979, 1986; Solon 1992, 1999). They postulate that higher educated parents will have higher incomes which will lead to increased investment in their children’s education, leading to better educational outcomes for their children. In this study we focus on understanding the long-term patterns and trends in the transmission of education across generations and how they relate to existing policies. We aim to provide measures of intergenerational transmission of education that are directly comparable with those provided by Hertz et al. (2007); hence, the methodology followed in this study will closely follow theirs. The transmission of educational attainment from parents to children across generations is modeled as a simple first-order Markov process as follows:

where the child’s educational attainment (S_c) is a linear function of parental education (S_p) in years of schooling. The constant term α captures all other factors that affect the child’s schooling, and ε_c is a mean zero error term independent of S_c and is independently distributed across generations and individuals. Hence, the estimated value of the coefficient β measures the persistence of educational attainment across generations. The higher the value of  β^  the greater the effect of parental education on children’s education, and vice versa. Equation [1] is estimated for each of the 10 birth cohorts.

A second measure of intergenerational persistence is the correlation between parental and child schooling, which gives a direct measure of the effect of parents’ schooling on that of their children (de Broucker and Underwood 1998; Hertz et al. 2007; Azam and Bhatt 2012; Chechi, Fiorio, and Leonardi 2013). The intergenerational correlation coefficient ρ is defined as ρ=β(σp/σc) where σp and σc are the standard deviations of parents’ and children’s schooling, respectively. The correlation coefficient ρ is therefore equivalent to the regression coefficient adjusted for the relative variation in education across generations and hence is a standardized measure of persistence and factors out the cross-sectional dispersion of education across generations.^[17] If the variance of the child’s education is lower than that of the parents, the correlation coefficient will exceed the regression coefficient and vice versa. If the variance of children’s schooling and parents’ schooling is identical, the regression parameter  β^  will equal the correlation coefficient ρ. Any changes in the relative variance will therefore result in both the correlation and regression coefficient behaving differently.

5 Trends in Intergenerational Mobility in Education: Empirical Results

We estimate the regression coefficients and correlations for parent–child education for each of the 5-year birth cohorts based on the number of years of education. Following Hertz et al. (2007), we derive the simple average across cohorts for Australia, which will render it directly comparable with their results for 42 countries. The average correlation coefficient for Australia is 0.31, which positions it as the second most mobile country, behind Denmark within the sample of OECD members (see Table 2). This is also comparable to the correlation coefficient for Australia reported by de Broucker and Underwood (1998) using a different dataset, but a similar age range.^[18]

Regression coefficients and correlations by birth cohort and indicate that intergenerational persistence of education has generally decreased over the last five decades, but the trend has not been consistent (Table 3). The regression coefficient has almost halved from 0.401 in the oldest cohort (1941–1946) to 0.203 for the youngest cohort (1987–1991). Moreover, the explanatory power of this relationship has declined indicating that educational outcomes of children depend less on parental education over the generations. The correlation coefficients follow a similar trend, declining from 0.381 for the oldest cohort to 0.24 for the youngest. There has been a significant improvement in intergenerational education mobility from the 1950s toward the end of the 1960s when both measures dropped significantly. The regression coefficient fell from 0.437 in 1947–1951 to 0.223 in 1962–1966 and the correlation coefficient fell from 0.443 to 0.238 during the same period. This trend has since slowed down with both measures ranging between 0.2 and 0.3 with the exception of the 1982–1986 cohort when the coefficient and correlation rose to 0.322 and 0.341, respectively.

In order to examine how these trends vary across regions within Australia, we estimate coefficients and correlations by state across all cohorts (1941–1991) and for two subperiods: a younger cohort born between 1976 and 1985, and an older cohort born between 1956 and 1965 which will provide a snapshot of regional variations in intergenerational education persistence over time (Table 4).^[19] While not all regional variations in the persistence of education can be explained by state-level policies, the trends are useful in informing how regional intergenerational mobility has changed over time. As evident from Table 4 there is considerable regional variation in mobility across the states. Tasmania had the lowest intergenerational educational mobility with a regression coefficient of 0.503 and a correlation coefficient of 0.439, while Western Australia had the highest with a regression coefficient of 0.221 and a correlation of 0.243. Across the two cohorts, we observe that intergenerational persistence of education has fallen in the states of Victoria, Western Australia and Tasmania, reflecting the declining dependency of children’s education on that of their parents’ education. On the other hand, persistence has increased in New South Wales, Queensland and South Australia.

These outcomes are partly a reflection of the evolution of the education system and policies over time, as well as the regional variations in socioeconomic and demographic trends. In Australia, the provision of education is the responsibility of the state governments with significant federal government funding. While there are national guidelines in place, the implementation of policies and funding decisions are determined at the state level which in principle could lead to variation in education policies across states.^[20] In addition, there are key historical and political differences in education policies across the states that would impact access to education and persistence across generations. For example, the state of Victoria has had a long history of promoting vocational and technical education which made secondary schooling accessible to middle class families. When the states of South Australia, New South Wales and Tasmania introduced secondary school fees in the 1930s, the Victorian government maintained free technical schooling which resulted in an increase in enrolments in these schools, even as enrolments in other states were declining (ABS 2001a). Studies on education policy suggest that regional polices and institutions play an important role in producing inequality in educational opportunities.^[21] Moreover, economic growth over time has not been uniformly distributed across the states in Australia with the resource rich states of Western Australia and Queensland outperforming the rest of the country in recent years.^[22] These regional differences can potentially contribute to variations in the socioeconomic and demographic composition of states. The heterogeneous nature of intergenerational education mobility over time across states is likely a combination of these and other factors which will be further explored in future extensions of this research.

Looking at the gender differences in these trends (Figure 2), there is a marked difference in the long-term trends in educational persistence between males and females. Both measures are higher for females indicating that a daughter’s education is more dependent on parental education compared to that of a son. Education mobility increased for males until the early 1980s, until it experienced a sharp decrease in 1986, before increasing again in the early 1990s. The long-term trend for women is different. Intergenerational education mobility of daughters increased dramatically between 1946 and 1966, with both measures falling from around 0.5 to 0.2 during this period. This trend is reversed in the early 1970s when both measures start inching up – until it tapered off in the 1980s. In keeping with existing evidence^[23] it appears that a daughter’s education continues to depend relatively more on that of her parents’ education than for a son.

Figure 2:

Intergenerational education correlations and coefficients by birth cohort: sons and daughters.

The period from 1960 to late 1970s saw increased government expenditure on education and major school reform, which were factors in increasing enrollment and retention in secondary schools (ABS, 2001b). University education was made free for a brief period from 1974 to 1988. Those born between the mid-1950s to the late 1960s were the beneficiaries of these policies, as they reached secondary school when access to schools was expanding and entered university when tuition was free. We observe this in the decreasing dependency of children’s education from their parents (Figure 2 and Table 3). As evident from Figure 2, girls have especially benefitted from these early expansions in education, with a significant increase in intergenerational education mobility for those born in the 1950s and 1960s which is also consistent with increasing female participation in schools (ABS, 2001b). The late 1970s and the 1980s saw a period of stagflation, and public expenditure on education fell. This, coupled with increased government funding toward the non-government schools sector, meant that children’s education was now again dependent on parental education (and income). During this period, education was conceived as expensive and the opportunity cost of education too high which would have disproportionately affected women. This is partly reflected in the rising correlations and coefficients for women during this period (Figure 2).

5.1 Education Transition Matrix and Mobility Indices

Next we examine the extent of education mobility conditional on the location of the parent along the education distribution using Markov transition probability matrices. The transition matrices are computed using the highest educational level attained by parents (generation t) and children (generation t+ 1) based on three educational outcomes (states) – secondary incomplete, secondary complete and post-secondary. Using the standard approach, let pij be the probability that educational outcomes (states) move from state i in generation t to state j in generation t+ 1, where i, j = 1,2,3 represent the three possible educational outcomes for parents and children, respectively. Then the observed transition probabilities are given by a 3 × 3 matrix P, with elements pij representing the probability of a parent with education level i having a child with level j education.

We estimate transition probability matrices for all children, daughters and sons for the whole period (1942–1991), and two subperiods: 1976–1985 (26 to 35-year-olds in 2011) and 1956–1965 (46 to 55-year-olds in 2011) for all parents and children (Table 5). Each row of the table gives the education level of the parent, while the columns indicate the education level of the child. Overall, the intergenerational persistence of educational attainment has fallen over time for both males and females, even though the level of dependency is higher for daughters. The probability of children whose parents had not completed secondary education, attaining a post-secondary qualification was 0.513 in the older cohort (birth years 1956–1965). This has increased to 0.596 for the younger cohort born between 1976 and 1985. Moreover, the probability of a child whose parents did not complete a secondary qualification, also not completing secondary education was 0.344 in the older cohort. This has fallen significantly in the younger cohort to 0.174.

Table 5:

Transition probability matrices and mobility indices of educational outcomes, for all, younger and older cohorts.

Parental education	Child’s education				Praise–Shorrocks mobility index
	Secondary incomplete	Secondary complete	Post-secondary	Total
Period 1942–1991: all children
Secondary incomplete	0.343	0.168	0.489	1.000
Secondary complete	0.160	0.274	0.566	1.000	0.838
Post-secondary	0.091	0.202	0.707	1.000
Period 1942–1991: sons
Secondary incomplete	0.301	0.169	0.530	1.000
Secondary complete	0.165	0.261	0.574	1.000	0.871
Post-secondary	0.085	0.219	0.696	1.000
Period 1942–1991: daughters
Secondary incomplete	0.376	0.167	0.457	1.000
Secondary complete	0.156	0.285	0.560	1.000	0.811
Post-secondary	0.096	0.186	0.718	1.000
Period 1976–1985 (younger): all children
Secondary incomplete	0.174	0.229	0.596	1.000
Secondary complete	0.121	0.220	0.659	1.000	0.904
Post-secondary	0.051	0.152	0.797	1.000
Period 1956–1965 (older): all children
Secondary incomplete	0.344	0.143	0.513	1.000
Secondary complete	0.194	0.211	0.595	1.000	0.835
Post-secondary	0.122	0.103	0.775	1.000

^[*]

5.1.1 Prais–Shorrocks Index, Upward/downward and Immobility Indicators

The Prais–Shorrocks mobility index (M_ps) is based on the transition matrix and illustrates the probability that an individual will stay in the same educational outcome as their parents. Let n be the number of educational outcomes (states) and pij the probabilities that are located on the main diagonal of the transition matrix. Then, M_ps is defined as follows (Shorrocks 1978; Prais 1955):

(2)Mps=1n−1∑j=1n(1−pjj)

M_ps usually takes values in the interval [0,1] with zero implying perfect immobility.^[24] We estimate the Prais–Shorrocks index for the whole sample and older and younger cohorts for all children and sons and daughters. The results are presented in the last column of Table 5 and shows that there has been significant improvement in education mobility over time. The M_ps increased from 0.835 for the older cohort (birth years 1956–1965) to 0.904 for the younger cohort born in 1976–1985. For the pooled sample (1941–1991), we find that a daughter’s education is relatively more dependent on her parents’ education than a son’s education. This confirms the trends we observe in Figure 2, where the correlations and coefficients are higher for daughters.

Extending the transition matrices, we construct three absolute mobility indicators: upward mobility (UM), downward mobility (DM) and immobility (IM). The immobility index is calculated as the average of the entries along the main diagonal of the transition probability matrix. The average of the entries above the main diagonal gives the downward mobility indicator, while the average of the entries below the main diagonal gives the upward mobility indicator. These mobility indicators are calculated for each 5-year birth cohort and presented in Table 6. The probability of upward mobility increased from 0.306 for the oldest cohort (1942–1951) to a high of 0.499 in the 1972–1976 cohort before tapering down to 0.326 for the youngest birth cohort (1987–1991). Immobility has remained relatively steady over time, declining slightly from 0.456 for the oldest cohort to 0.419 for the youngest cohort. Downward mobility has also remained relatively stable for the most part until it started rising for the two youngest cohorts. The last two rows compare the average over the entire period (1946–1991) for each of the mobility indicators with OECD figures for Australia which cover the period 1947–1987^[25] and show that they are broadly comparable.

Table 6:

Upward mobility, downward mobility and immobility ratio of children’s education outcomes by birth cohort.

Birth cohort	UM	DM	IM	Sum
1942–1946	0.306	0.238	0.456	1.000
1947–1951	0.357	0.142	0.501	1.000
1952–1956	0.402	0.151	0.447	1.000
1957–1961	0.425	0.145	0.430	1.000
1962–1966	0.404	0.134	0.462	1.000
1967–1971	0.427	0.152	0.421	1.000
1972–1976	0.499	0.096	0.406	1.000
1977–1981	0.497	0.091	0.412	1.000
1982–1986	0.464	0.141	0.395	1.000
1987–1991	0.326	0.255	0.419	1.000
Average 1946–1991	0.408	0.151	0.442	1.000
Average 1947–1987a	0.41	0.13	0.46	1.000

Source: HILDA survey 2011.

^[*]

5.1.2 Relative Educational Opportunities

We compute relative educational opportunities indicators that show the extent to which the observed educational attainment of children are distributed across parental education. Two conditional probability measures are derived for different levels of education: post-secondary (R_ps) and secondary (R_s). R_ps is the probability that a child has completed post-secondary (ps) level of education, conditional on parents attaining a post-secondary education, relative to the probability that a child has completed post-secondary education, given that the parents have not completed secondary education (si). That is,

(3)Rps=probchildps|parentpsprobchildps|parentsi

R_s is the probability that a child has completed secondary level of education, conditional on parents attaining a secondary education, relative to the probability that a child has completed secondary education, given that the parents have not completed secondary education. This is defined by

(4)Rs=probchildsc|parentscprobchildsc|parentsi

These relative educational opportunity measures computed for each 5-year birth cohort are presented in Figure 3. The ratio R_ps has declined gradually over time, even though it has remained relatively stable from around the early 1950s to the early 1980s. The chance of a child born in the oldest (1942–1946) cohort to parents with post-secondary education, completing a post-secondary education is a factor of 1.62 higher than that of a child whose parents did not complete a secondary education. This has fallen to 1.19 for the youngest cohort born between 1987 and 1991. At the secondary level, R_s declined sharply between the 1940s to the early 1960s and more gradually since then. A child born in the oldest cohort (1942–1946) to parents with a secondary education had a 2.72 higher chance of completing a secondary education, compared with those with parents who did not complete a secondary education. This was 1.57 for a child born in the youngest cohort (1987–1991). Moreover, R_ps is higher than R_s between the mid-1960s to the early 1980s suggesting that the impact of parental education was high at the post-secondary level during this period.

Figure 3:

Relative educational opportunities by birth cohort.

6 Conclusions

The Australian education system has gone through a range of changes during the last five decades, from changes in financing of education to increasing participation. This study investigates the long term trends in intergenerational transmission of education in Australia using data from the 2011 HILDA survey to construct a range of measures including regression coefficients and education correlations between parents and children, as well as estimating a set of mobility indices. Our overall results indicate that there has been a significant increase in intergenerational educational mobility during the last five decades which positions Australia as a high mobility country among OECD members. There are, however, considerable regional differences in the persistence of education across generations which are masked at the national level. Daughters’ education continues to be relatively highly correlated with their parents, compared to sons. Moreover, while the relative probabilities of children from low education parental background achieving high educational levels compared to high education parents have improved over time, the changes at lower levels of education are marginal. Future extensions will focus on identifying causal relationships that explain the trends found in the current study.

From a policy perspective, the expansionary education policies in the post-war period have paid off. Education attainment, participation, especially of girls, and retention increased dramatically, with the resulting increase in intergenerational education mobility over time. However, the slowing down, and even reversing in some instances, of this trend in the recent years pose some important questions on the distributional and equity outcomes of subsequent policies. The impacts of the proposed policy changes in the 2014 Australian federal government budget are yet to be seen, and the findings in this study will contribute to the ongoing debate on the equity of the education system.