02.06.2014 Views

Evaluation of the Australian Wage Subsidy Special Youth ...

Evaluation of the Australian Wage Subsidy Special Youth ...

Evaluation of the Australian Wage Subsidy Special Youth ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

87<br />

was an issue for <strong>the</strong> data. Unlike <strong>the</strong> Baker (1984) analysis, survey non-response was<br />

ignored by Rao and Jones (1986).<br />

2.3.4 Richardson (1998)<br />

Richardson (1998) used panel data spanning 4 years <strong>of</strong> repeat surveys <strong>of</strong> <strong>the</strong> <strong>Australian</strong><br />

Longitudinal Survey (ALS). The analysis is based upon <strong>the</strong> ALS list sample, which was a<br />

sample selected from <strong>the</strong> CES records for young unemployed aged 15-24. The ALS list<br />

sample was a nationally representative sample <strong>of</strong> <strong>Australian</strong> youths aged 15-24 who had<br />

been registered as unemployed with <strong>the</strong> Commonwealth Employment Service for at least<br />

3 months in June 1984. They were interviewed in September-October each year from<br />

1984-1987. 58 Using <strong>the</strong> labour market history, those who entered SYETP job placements<br />

during 1984 until <strong>the</strong> 1985 interview defined <strong>the</strong> SYETP treatment group and comparison<br />

group reference period. The subsequent survey data on labour market history until <strong>the</strong><br />

1987 interview date was used to examine post-programme employment outcomes.<br />

An analysis <strong>of</strong> attrition was made by examination <strong>of</strong> <strong>the</strong> means. It was concluded that <strong>the</strong><br />

characteristics were virtually identical and so attrition bias was treated as a minor<br />

problem (Richardson (1998): 5). It was however noted that those lost to attrition were<br />

more likely to have less than year 10 education, more likely to have held a job for at least<br />

1 year and less likely to have never held a job. Accordingly, no treatment <strong>of</strong> non-response<br />

was made. It was noted that <strong>the</strong> treated SYETP had different characteristics than <strong>the</strong><br />

comparison group <strong>of</strong> non-participants. It was observed that in <strong>the</strong> raw data, employment<br />

in 1986 was 14 percentage points higher for those who had participated in SYETP than<br />

<strong>the</strong> comparison group, and in 1987 it was 5 per cent higher than <strong>the</strong> comparison group.<br />

First, separate probit models <strong>of</strong> employment and <strong>of</strong> participation were estimated. Then<br />

Richardson used a Heckman bivariate probit which controlled for programme selection<br />

by modelling both <strong>the</strong> participation and post-programme employment. To identify <strong>the</strong><br />

model, a variable showing CES referrals to ano<strong>the</strong>r programme, and age <strong>of</strong> <strong>the</strong><br />

58 Fur<strong>the</strong>r details <strong>of</strong> <strong>the</strong> data are in <strong>the</strong> later chapter dealing with <strong>the</strong> replication study.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!