This report was originally published 19 January 2016.

Summary

The NSW Government Department of Education and Communities has commissioned the University of Wollongong to conduct research on the role of education in intergenerational mobility in Australia. This is the revised Final Report on the project.

The economic literature on this question is not large, especially for Australia. One strand of the empirical literature has sought to estimate causal effects of education, mainly exploiting compulsory schooling reforms as natural experiments. A second strand examines the extent to which geographical variations in intergenerational mobility can be explained by differences in the characteristics of education systems. A third strand of the literature has explored the role of education as a mediator – that is, the role of education as a pathway through which family background affects economic outcomes in the next generation. Our study is in this third strand. The main objective of our research is to study the extent to which education is a mechanism which explains intergenerational transmission of economic (dis)advantage in Australia.

To this end, we have developed an approach which seems to be a methodological innovation. Our approach is related to a mediation model (Baron and Kenny, 1986), but is better able to directly account for the range of family background characteristics which may affect child earnings through the pathway of education or through other mechanisms. The approach estimates the extent to which education mediates the combined effects of all such background variables on child earnings. This methodological innovation has been scrutinised at several academic seminars and conferences, leading to several refinements. But it has not yet been subjected to a formal academic peer review process and so the findings in this report should be treated with caution.

The analysis suggests that family background is a much stronger determinant of economic wellbeing than implied by previous studies:

• Using conventional techniques with the latest available data, we estimate that the ‘intergenerational elasticity’ of wages is 0.35, which is 34% higher than implied by influential previous studies.
• When all available family background data are incorporated (not just parental earnings), the importance of background is more apparent. For example, males at the 90th percentile of the ‘family background index’ have expected earnings that are 65% higher than those at the 10th percentile. For females, the corresponding difference is also large (53%). Similarly, children from low socioeconomic backgrounds are unlikely to have high earnings themselves. Amongst people from the lowest quartile of family background, around 40% themselves have earnings in the lowest quartile, while only about 12% have earnings in the top quartile.

There is a positive relationship between socioeconomic background and education, and a positive relationship between education and earnings. It follows that education is one of the mechanisms through which economic advantage is transferred from one generation to the next. As a guiding principle to aid interpretation of the main results, it is noted that if education explains a ‘small’ component of intergenerational transmission, this implies that access to education is relatively equitable – i.e. family background is not a strong determinant of educational attainment.

The main analysis was conducted using the 2012 wave of HILDA, Australia’s main household panel survey. The results suggest that:

• Own-education accounts for around 25%-40% of intergenerational transmission of economic advantage in Australia.
• The role of education appears to be larger for females than for males. There are several potential explanations for this, but it may be explained by a larger ‘direct’ effect (i.e. through mechanisms other than child’s education) of family background for males.
• The mediating role of education is largest for the parental education component of family background.

We have used another, complementary, approach to consider the role of the education system in intergenerational transmission. In a parallel analysis, we estimated the effect of family background on educational attainment, and found it to be considerably (30%) smaller than the effect of family background on earnings. This result provides a different perspective on the extent to which the education system is ‘part of the solution’ rather than ‘part of the cause’ of intergenerational transmission of wellbeing.

We also conducted a complementary analysis of the British Cohort Survey data:

• The analysis suggests that the role of education in transmitting economic advantage is similar in the UK to that of Australia.
• The quality of data items measuring skills in HILDA is relatively poor, and these are not measured in childhood. The British Cohort Study includes high quality items on child’s cognitive and non-cognitive skills. Nevertheless, in a ‘comparable’ analysis of BCS, the estimated role of skills is similar to that found in HILDA. Therefore the lower quality skills measures in HILDA do not seem to bias the Australian results.

This report presents a ‘big picture’ view on the role of education in intergenerational economic mobility in Australia. The project has not addressed causal questions on the extent to which educational programs or interventions can lift people out of economic disadvantage. The most credible research on such questions has used quasi-experimental techniques that exploit policy changes such as compulsory schooling laws. Rigorous impact evaluations of smaller programs may also be a fruitful avenue for further research. Finally, incorporating elements of random assignment into trials of new initiatives is likely to yield the highest quality evidence on their causal effects on student outcomes, including for students from disadvantaged backgrounds.