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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

89<br />

Table 2.21 Richardson (1998) Estimated marginal effect <strong>of</strong> SYETP on employment from<br />

bivariate probit modelling<br />

effect on employment 1986 effect on employment 1987<br />

All data used 26.4 19.7<br />

Excluding jobs or placements<br />

retained to 1986<br />

23.7<br />

Source: Richardson (1998) p22 Table 6, and p12.<br />

The Richardson (1998) analysis usefully modelled both <strong>the</strong> participation in SYETP and<br />

<strong>the</strong> post-programme employment using <strong>the</strong> Heckman selection method in a bivariate<br />

probit. Earlier SYETP evaluations only modelled employment, none modelled<br />

participation in SYETP. The modelling included numerous variables that account for<br />

individual characteristics. This was by far <strong>the</strong> largest set <strong>of</strong> individual characteristics used<br />

in modelling <strong>the</strong> post-programme employment for SYETP. Generally, most <strong>of</strong> <strong>the</strong>se have<br />

been used in <strong>the</strong> economics literature, if available in <strong>the</strong> data, to explain employment,<br />

education, or programme participation. The wide variety <strong>of</strong> variables included illustrates<br />

<strong>the</strong> very rich dataset <strong>the</strong> ALS has to draw on. The inclusion <strong>of</strong> a wide set <strong>of</strong> individual<br />

characteristics can help demonstrate that all <strong>the</strong> useful observable characteristics have<br />

been controlled for in modelling both <strong>the</strong> employment and programme participation.<br />

However some <strong>of</strong> <strong>the</strong> variables, such as religion, are possibly more unusual for<br />

describing economic models <strong>of</strong> employment or programme participation.<br />

Although <strong>the</strong> non-response was briefly examined, <strong>the</strong> examination was terse. The<br />

conclusions about <strong>the</strong> possible effects <strong>of</strong> non-response were based on little serious<br />

contemplation <strong>of</strong> non-response. Some earlier SYETP evaluations (Stretton (1984), Baker<br />

(1984)) modelled non-response for <strong>the</strong>ir observational data and found non-response was a<br />

potential problem. Richardson (1998) contains no modelling <strong>of</strong> non-response, yet <strong>the</strong><br />

potential for non-response is much higher in a panel constructed from 4 surveys. As for<br />

earlier SYETP evaluations, again no treatment for non-response was made, although in<br />

<strong>the</strong> case <strong>of</strong> Richardson (1998) this was in line with <strong>the</strong> conclusions whereas for earlier<br />

evaluations this was in spite <strong>of</strong> <strong>the</strong> conclusions about <strong>the</strong> existence <strong>of</strong> a potential nonresponse<br />

problem for estimates.

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

Saved successfully!

Ooh no, something went wrong!