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.

93<br />

Controlling for differences in individual characteristics between <strong>the</strong> comparison and<br />

treatment groups was carried out in all <strong>of</strong> <strong>the</strong> SYETP evaluations using econometric<br />

modelling <strong>of</strong> <strong>the</strong> post-programme employment. This accounts for observed differences in<br />

individual characteristics between <strong>the</strong> comparison and treatment groups by including<br />

<strong>the</strong>m in <strong>the</strong> model. The earlier modelling showed that controlling for individual<br />

characteristics was important, but <strong>the</strong> set <strong>of</strong> characteristics was limited. The key issue<br />

<strong>the</strong>n is whe<strong>the</strong>r all <strong>the</strong> observable characteristics that might affect employment have been<br />

included. The Richardson (1998) analysis had a very broad set <strong>of</strong> characteristics and it<br />

showed that some <strong>of</strong> <strong>the</strong>se, such as education, are indeed important for modelling<br />

employment. Leaving out important variables for modelling employment is likely to lead<br />

to specification bias.<br />

Controlling for differences in individual characteristics between <strong>the</strong> control and treatment<br />

groups is treated fur<strong>the</strong>r in this study when different modelling methods are used. As <strong>the</strong><br />

earlier review <strong>of</strong> wage subsidy evidence shows, matching techniques can also be used to<br />

resolve differences in individual characteristics between <strong>the</strong> comparison and treatment<br />

groups. Later <strong>Australian</strong> evaluation <strong>of</strong> <strong>the</strong> Jobstart wage subsidy used direct matching<br />

techniques. These methods however require a limited set <strong>of</strong> characteristics and large data<br />

sample in order to enable <strong>the</strong> analysis to proceed. The more characteristics it is desirable<br />

to match on, <strong>the</strong> fewer matches become available amongst <strong>the</strong> treatment and comparison<br />

groups. As <strong>the</strong> earlier modelling <strong>of</strong> SYETP showed, controlling for individual<br />

characteristics was important, and <strong>the</strong> Richardson (1998) evidence indicates <strong>the</strong> breadth<br />

<strong>of</strong> characteristics that might be useful. Matching, such as that <strong>of</strong> <strong>the</strong> Jobstart evaluations,<br />

which does not control for <strong>the</strong>se o<strong>the</strong>r characteristics that affect both employment and<br />

participation would <strong>the</strong>n be subject to bias. As such, it is <strong>the</strong>n most useful to control for a<br />

fairly broad set <strong>of</strong> characteristics. O<strong>the</strong>r recent overseas evidence has used propensity<br />

score matching (PSM) techniques, which allow a greater number <strong>of</strong> characteristics to be<br />

controlled for, for example Bonjour et al (2001). This method has not been used for any<br />

published <strong>Australian</strong> evaluation so far. The PSM method is applied in <strong>the</strong> following study.

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

Saved successfully!

Ooh no, something went wrong!