The Australian Economic Review, vol. 34, no. 2, pp. 135–54  The University of Melbourne, Melbourne Institute of Applied Economic and Social Research 2001 Published by Blackwell Publishers Ltd, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA * The research was partially funded through a fellowship from the Ronald Henderson Research Foundation. We are indebted to Jon Altman, Tom Crossley, Yi-Ping Tseng, Will Sanders and John Taylor for comments on an earlier draft. The views expressed in this article do not necessarily reflect those of the Australian Institute of Family Studies. Abstract Data from the 1986, 1991 and 1996 censuses are used to conduct a synthetic cohort analysis of the income distributions for Indigenous and non-Indigenous males and females. The advan- tage of this approach is that statistical tech- niques can be used to control for unobservable differences between the Indigenous and non- Indigenous populations, such as ability and schooling quality, as well as assimilation, dis- crimination and other attitudes. The results demonstrate that the failure to control for un- observed differences in existing studies of In- digenous income will induce a significant bias in both empirical and policy analysis. Trends in relative income are also identified and are related to broad changes in labour force status. The deliberate policy shift in the early 1990s to paying welfare to individuals (‘individualisa- tion’) has resulted in an increase in financial independence among many females. The other insight from the analysis is that the generosity of welfare payments or improved targeting of benefits has materially advantaged extremely poor Indigenous people. While this is a positive outcome in its own right, policy also needs to take into account the interaction between tax, welfare, productivity and incentives to work. 1. Introduction The myth of equality in Australian society is clearly exposed by the large income gap be- tween Indigenous and other Australians. Tay- lor and Hunter (1998) estimate, in aggregate terms, that Indigenous incomes would have to increase by $1.6 billion per annum (in 1996 dollars) to achieve income equality. This num- ber is remarkable given that Indigenous people only constitute about 2 per cent of Australia’s population. This article describes this income gap in detail and teases out the reasons for the low levels of income experienced by many In- digenous Australians. The unit of analysis used is a ‘cohort’ which can be uniquely identified in successive cen- suses by its sex, age and Indigenous status. Rather than mechanically updating previous income data, this article sheds new light on the income deprivation of Indigenous Australians by tracing changes in income over the life cycle for Indigenous and non-Indigenous cohorts. This cohort analysis provides the first ‘longitu- dinal’ data on Indigenous income. Whilst there is no pretence in this article to be following the same individuals across time, Deaton (1985) describes how carefully constructed cohorts permit a ‘synthetic panel’ or ‘longitudinal’ analysis. The census data on income indicate pre-tax income from all sources rather than separately identifying sources of income. Census income therefore reflects a complex interaction be- tween age, the social security and transfer pay- ment system, labour income, assets and other Analysing Recent Changes in Indigenous and Non-Indigenous Australians’ Income: A Synthetic Panel Approach B. H. Hunter and M. C. Gray* Centre for Aboriginal Economic Policy Research, The Australian National University and Australian Institute of Family Studies, respectively