The Contribution of Structural Shocks to Australian Unemployment* CHRIS HEATON Macquarie University, Sydney, Australia PAUL OSLINGTON Australian Defence Force Academy/University of New South Wales, Canberra, Australia In this paper dynamic factor analysis techniques are used to decompose changes in unemployment into industry sectoral and common components. Sectoral shocks are important, but the dominant causes of variation in unemployment are common to all industries. This is particularly the case for low-frequency fluctuations in unem- ployment. The pattern of the estimated sectoral shocks reflects the well-documented shift of employment from agriculture and manufac- turing to services, and we find no evidence that microeconomic reform has contributed greatly to unemployment. I Introduction Structural unemployment has been widely dis- cussed in the academic and popular literatures, but there is little consensus about how much of the rise in unemployment since the 1970s can be attributed to structural factors. The issue is of considerable importance for public policy because, if structural factors are dominant, emphasis should be placed on industry, labour market, and other microeconomic policies rather than macroeconomic policies in fighting unem- ployment (Productivity Commission 1997, 1998). Most Australian work on structural unemploy- ment has used techniques from an important paper by Lilien (1982) on the contribution of industry sectoral shocks to USA unemployment movements. 1 Lilien constructed an index of the volatility of employment growth rates across 11 USA industries for the period 1948–1980, and added the index as a variable to a macroeconomic model in which changes in unemployment are driven by unanticipated monetary shocks. Find- ing the volatility index variable was statistically significant, he decomposed unemployment move- ments into a macroeconomic component (attrib- utable to the unanticipated monetary shocks) and a sectoral component (attributable to the volatil- ity index). This decomposition also yielded a natural rate of unemployment series, where the * Earlier versions were presented at the University of Melbourne, Australian National University, University of Oxford and European Meetings of the Econometric Society. We thank participants for many helpful comments, and also the anonymous referees. Correspondence: Paul Oslington, School of Eco- nomics and Management, University of New South Wales/Australian Defence Force Academy, Northcott Drive, Canberra ACT 2600, Australia. Email: p.oslington@adfa.edu.au 1 Lilien (1982) and much of the related literature focuses on the industry sectoral dimension of structure. There exist other dimensions of structure (e.g., occupation, age, gender) which might also be considered. The methods used in this paper can be extended to analyse several dimensions of structure simultaneously. However, while the Australian Labour Market Survey should yield the cross-tabulated unemployment data that we would need to extend our study to other dimensions of structure, the Australian BureauofStatisticswasunabletosupplysuchdataafter lengthynegotiations.Accordingly,wefollowmostofthe literature in considering only the industry sectoral dimension of structure. 433 Ó 2002. The Economic Society of Australia. ISSN 0013–0249. THE ECONOMIC RECORD, VOL. 78, NO. 243, DECEMBER, 2002, 433–442