The government is now trying to spin the Work Programme figures, as expected, by focusing on the initiative's "cost effectiveness". The BBC's Nick Robinson, for instance, writes:
Ministers claim that they are meeting their "off benefit targets" and that they are saving money too. The cost of every job secured under their Work Programme is, they say, just over £2,000 compared with a cost of almost £7,500 under Labour's [Flexible] New Deal because the contractors are only paid 60% of their fee once someone is in a sustainable job: ie for six months.
It's certainly the case that Labour's programmes were more expensive than the coalition's replacements. But what this spin demonstrates is a serious failure to control for background noise. The Work Programme is, so far, worse than nothing at ensuring "job outcomes" – that is, people in unsubsidised work six months after they leave the programme. In the first fourteen months, 3.5 per cent of participants achieved job outcomes, but for people not on the programme, 5 per cent were expected to get jobs, according to Labour's shadow minister Liam Byrne.
(The news shouldn't be hugely surprising – one very effective way to get a job is to spend all day every day applying for jobs. Any training programme has to overcome that hurdle.)
Some quick back of the envelope maths, here. The full data is simply not available, but if ministers are saying that the Work Programme cost £2000 per job, and we know that there have been 32,310 job outcomes, then presumably they are claiming a budget to date of £65m.
Given that 5 per cent background rate, we can expect that if the Work Programme had never been instituted, there would have been 46,000 jobs in the normal process: 14,000 more.
In other words, the Work Programme did not cost £2000 per job. Instead, for every £4,600 it spent, it destroyed one participant's chance of employment.
Updated: The effect of the work programme was on the 14,000 job difference, and so the effect is one job destroyed for every £4,600, not for every £1,400. 3.5 per cent is the result for the first fourteen months, not the first year. Clarified the source of the 5 per cent figure.