Indigenous data from the ABS Labour Force Survey: what can they tell us?

The Australian Bureau of Statistics (ABS) has recently published annual estimates of Indigenous labour force status for the period 1994-2000 based on the Labour Force Survey (LFS). Considerable interest surrounds these for two reasons: · as annual estimates, they offer the possibility of establishing trends in labour force status that are closely aligned-at least more so than are census data-with identifiable policy shifts and macroeconomic shocks; · the time series charted from the sequence of survey results is suggestive, at least at face value, of a sizeable decline in the Indigenous unemployment rate since the mid 1990s, and a current upward trend in employment levels. This paper critically evaluates these estimates with a view to assessing their utility for policy evaluation. The estimates are first examined in the context of previous attempts to benchmark the position of Indigenous people in relation to the labour market. The LFS methodology is then evaluated (to the extent that it bears on the interpretation of results), and finally the standard errors associated with annual movement in the estimates are calculated. Along the way a series of policy questions are addressed: do these apparently positive results suggest that improvement in the position of Indigenous people in the labour market is at long last emerging; · do they reflect success of the Indigenous Employment Policy (IEP); · do they result from macroeconomic or microeconomic change? Findings The LFS experimental estimates are of limited value for policy analysis. At best, they merely confirm existing understandings of recent labour force trends. At worst, they are unreliable due to high standard errors. In particular, movements of annual rates are statistically insignificant in all but the last two years, thus preventing the establishment of long-term trends. The key finding of significant decline in unemployment rates since 1998 resonates with an analysis of trends in Community Development Employment Projects (CDEP) scheme employment and with the fact that purely administrative changes to the scheme are likely to have raised overall employment levels over the same period. The prospect that recent decline in Indigenous unemployment has formed part of the general labour market trend appears unlikely. Also, the implementation of IEP would seem to have occurred too recently to have had any bearing on this result. The proposal by the ABS to augment the Indigenous sample in the LFS by providing annualised estimates is an innovative option, but may not be as straightforward as first appears. Such a procedure would require complex transformations of LFS data, probably involving out-of-sample estimates of the transition between, and duration of, respective labour force states. These are likely to effect the reliability of the final estimates.