This morning the Telegraph and Mail ran stories claiming the the government’s benefit cap was already proving a success nine months before it comes into effect. According to the Telegraph :
Iain Duncan Smith, the Work and Pensions Secretary, is due to release figures which show that 1,700 people who would have been affected by the £26,000-a-year limit have taken up work since being warned about next year’s cap ...
"These figures show the benefit cap is already a success and is actively encouraging people back to work," Mr Duncan Smith said. "We need a welfare state that acts as a safety net and encourages people back to work." Mr Duncan Smith said that the figures would embarrass Labour, which had opposed the cap.
The statistics on which the stories were based were released by DWP this morning after the press stories had appeared, a form of sharp practice for which they have already been ticked off by the UK Statistics Authority . Even had Labour opposed the benefit cap (unfortunately, they didn’t), there would be little for them to worry about in today’s figures, which should rather be an embarrassment to the government and to the gullible journalists who faithfully wrote up what they had been briefed. In fact, the data shows roughly the opposite of what Mr Duncan Smith claims.
The figures are based on contact made by JobcentrePLus with 58,000 claimants who it was believed would be affected by the cap when it comes into effect, assuming they were still claiming at that point. Over the two month period since letters were sent to affected claimants warning them of the policy change, 1,700 are said to have moved into work. That’s 2.9 per cent of the total.
But the obvious question seems not to have been asked: how many would have moved into work in any case?
We can get an idea from data on benefit flows. These are a lot higher than is usually realised: even in this period of weak labour demand, 89 per cent of claims for Jobseekers' Allowance and 73 per cent of claims for Employment Support Allowance end within a year  (pdf). But surely claimants receiving payments high enough to hit the cap spend longer on benefit? In fact, there’s no evidence for this, as the table shows.
|Duration on benefit as percentage of caseload||All out of work||Subject to cap|
|Up to six months||23||19|
|Six months up to one year||11||12|
|One year and up to two years||11||14|
|Two years and up to five years||16||23|
|Five years and over||40||32|
Source: Nomis and Commons Hansard 
The main contribution to benefit entitlement exceeding the cap level of £26,000 a year pro rata is high housing benefit payments. The average monthly off-flow rate from housing benefit over the last year was 2 per cent. If we take this as a proxy for people moving into employment, then over a two month period, other things being equal, we would have expected about 2,300 out of 58,000 people (4 per cent) to have taken up work. So an off-flow into employment of 1,700 is no indication whatsoever that the cap is affecting behaviour. The government is claiming this figure as a "success", when all it shows is that people receiving high housing benefit payments sometimes move into employment. Who knew?
I don’t think Duncan Smith is being disingenuous here. I fear it is much worse than that: he is genuinely self-deceived. If he thinks that an off-flow of this scale offers any evidence of the effect of policy, it is because he and his government are fixated on long-term benefit claimants, largely for ideological reasons.
Thus the fact that people actually leave benefits in very large numbers every month without being forced is routinely airbrushed out of the presentation of government policy, while ministers make ludicrous claims about "families where nobody has worked for three generations" (a misleading claim addressed by Lindsey Macmillan and Paul Gregg ).
So I suspect that the ideological message has been so profoundly internalised that the Secretary of State simply cannot conceive that anyone on this level of benefits could move into work other than in response to the threat of compulsion from his department, so any off-flow must count as evidence that the policy is succeeding.
Of course, I could be wrong. Maybe Duncan Smith is being disingenuous after all and knew exactly what he was doing when he sold the Telegraph and Mail this particular pup. That might even be less disturbing than the thought that he really believes this stuff.