After reading the Deloitte Access Economics report on the Skilling Queenslanders for Work (SQW) program, I have significant doubts about whether the program generated the large benefits claimed. Earlier today, the Brisbane Times reported (Jobseeker program axed, then praised):
Queensland’s employment minister axed a program a week before his department received a report praising the scheme to help disadvantaged jobseekers…
…The SQW program, introduced in 2007, delivered targeted grants-based labour market programs that aimed to reduce unemployment and under-employment among disadvantaged groups.
A report by Deloitte Access Economics said of the 57,000 persons who gained employment through SQW, 8500 of these would not have gained employment without the program. It said about $375 million in wages would be generated by these 8500 persons in 2012-13.
I suspect the 8,500 figure is an over-estimate of the number of people who gained employment due to the axed SQW program and otherwise wouldn’t have. The study estimates this number based on comparing the “treatment group” of SQW participants with a “control group” of jobseekers who participated in Commonwealth employment programs but not SQW. But the comparison is questionable, because when surveyed the SQW participants had had a much longer period in which to have found a job, meaning their employment outcomes would have been better. As explained on p. 83 of the Deloitte report:
The details on the employment pathways of the treatment group come from the OESR survey of SQW participants, conducted 12 to 18 months after the completion of their involvement with SQW…
…The details of the employment pathways of the control group come from the DEEWR Labour Market Assistance Outcomes (LMAO) reports, conducted 3 months after participation in Commonwealth employment assistance. [emphasis added]
A person is much more likely to have found a job after 12 to 18 months than three months, either because they eventually get lucky or lower their expectations. I think the report is incorrect to assert that:
While not perfectly aligned in terms of the period between participation and outcome measurement, any bias in the data could be argued to skew the results either in favour of SQW or against SQW.
In my view, the bias caused by the time difference between the program and the survey will skew the results in favour of SQW. Hence the Government shouldn’t be too concerned it didn’t see the report prior to axing the program.