1504.0 - Methodological News, Sep 2002  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 20/11/2002   
   Page tools: Print Print Page Print all pages in this productPrint All

A COHORT ANALYSIS OF UNEMPLOYMENT RATES FOR AUSTRALIA

Over the past 20 years the Australian labour market has experienced significant changes. The structure and operation of the labour market is substantially different today compared to 1981 due to changes in demography, technology, industrial relations and education patterns. All of these factors will have had an affect on the unemployment rate.

Relatively high rates of unemployment throughout the 1980s and 90s, when compared to rates in the 1960s and 70s, have prompted extensive research into the causes and effects of unemployment. Most Australian studies of unemployment at the micro level use cross sectional surveys as the basis of their analysis. Cross sectional surveys provide a snapshot of the labour market at a particular point in time. This enables comparison in a number of dimensions such as gender, age, household type and education. However, these comparisons do not reveal if the same people are unemployed in both periods or if some have found employment and been replaced by people newly unemployed.

Ideally, one would like to have panel data (or longitudinal survey data) to answer these types of questions. Panel data is where an individual can be tracked over time. In the absence of panel data one can use cross section data sets to undertake cohort analysis. Cohort analysis cannot provide insights into labour dynamics at the individual level but it can at a group or cohort level. Particular cohorts may experience relatively higher levels of unemployment over time than the rest of the population. For example, cohorts which enter the labour force during an economic downturn may be scarred by being unable to find employment for an extended duration and have relatively high unemployment rates throughout the rest of their lives as a result.

This study employed cohort analysis to examine patterns in unemployment rates for the period 1981 to 2001 using quasi-panel data constructed from the Labour Force Surveys. Regression techniques were used to decompose the unemployment rates into age, year and cohort effects. The cohort effects were further closely examined among males and females by looking at some driving forces that affect unemployment rates such as labour supply (participation), education levels and the conditions on entry into the labour market. The Analysis branch project team (Ravi Ravindiran, Terry Rawnsley and Annette Jose) found that cohort effects were a significant determinant of unemployment for men, along with age and year effects. But cohort effects were insignificant in explaining unemployment patterns among women.

Results of this study are geared towards expanding our understanding of the methodology underlying cohort analysis, and to contribute to the body of analytical tools in analysing existing ABS datasets.

For more information, please contact Ravi Ravindiran on (02) 6252 7039, or ravi.ravindiran@abs.gov.au.