Thursday, January 09, 2014



Use for the OIA: Uncovering piss-poor policy

Back in 2012, the government announced new rules requiring beneficiaries to pass pre-employment drug-tests. The rules came into force in July last year, and have been in force for six months now. But have they been effective? Someone used FYI, the public OIA requests site, in an effort to find out, asking the sort of basic questions you'd use to assess the effectiveness of the policy: how much has it cost? many tests have been passed? How many failed? How many of the failures were deemed to be addicts (and thus not subject to sanctions), and how many "recreational users" (who were)? Above all, how much money has the policy saved?

The answer? WINZ doesn't know. They have referred 8,100 beneficiaries to jobs requiring pre-employment drug screening, but have no idea how many of them passed their tests. They do know that 22 of them have "failed" (11 actual failures, and 11 refusals to piss into a cup so the Minister can get "tough on poor" votes), but they don't know how many of them have were addicts (and won't bother looking, even for such a small number). As for the amount saved, that part of the request is "refused under section 18(e) of the Official Information Act as this information does not exist". Estimates of the amount saved are also refused, but under s9(2)(g)(i), protecting free and frank expressions of opinion between officials - meaning that it would make the Minister look bad (which should not be any sort of reason for refusal).

So basicly WINZ isn't bothering to track information used to assess the effectiveness of the policy because they know its a showboat which will be assessed not on its merits in helping beneficiaries get jobs, but on how it makes the government look to voters. Which is a piss-poor way to make policy. Still, there's some useful information in there, in that 8,100 beneficiaries were referred to jobs requiring testing, while only 22 failed. If those "required" tests were actually demanded, then that implies a failure rate of 0.27%, so low as to be negligible. Whether we should be spending millions of dollars targeting this non-problem is left as an exercise for the reader.