Sinning with statistics: Frank Field on welfare and immigration

There is no question of Frank Field lying: but there are other ways to sin with statistics - as Mr Field proves through his analysis of benefit payments.

Frank Field

Frank Field writes in today’s Telegraph:

“It is the great paradox of welfare. When Tony Blair won the 1997 election, the total number of benefit claimants of working age stood at 5.7 million.

“When Gordon Brown went to the country in 2010, the level was the same — even though more than three million jobs had been created under Labour. The problem was that, of those new jobs, 80 per cent went to immigrant workers.”

In one of Dashiell Hammett’s crime stories, the detective finds himself in Tijuana staring at a sign over a bar that says:

“Only genuine pre-war American and British whiskeys served here’ trying to work out ‘how many lies could be found in those nine words.”

The quotation above evokes a similar response. Just how much inaccuracy and distortion can be squeezed out of a few apparently unambiguous statistics?

Of course there is no question of the saintly Frank Field lying: but there are other ways of sinning with statistics.

In this case, Mr Field has taken figures for total working-age benefit receipt for 1997 and 2010 to make a far-fetched claim about employment and migration, which can be summed up as “Benefit receipt didn’t change, so migrants took all the extra jobs”.

Now the number of working age claimants in May 1997 and May 2010 are indeed very close, 5.65m compared to 5,73m. But of course the figures for May 2010 reflect the impact of the recession: in 2007 for example the total number of claimants was just over 5m.

The blue and green curves both show numbers of out of work benefit claimants, measured on the left hand scale. The green curve excludes claimants of DLA. The dotted lines show the corresponding rates of receipt ie claimants as a percentage of the working age population. The data doesn’t correspond precisely to the figures cited in the article for various reasons. Data is from the Department of Work and Pensions’s (DWP) 5 per cent sample (1995-2002) and Work and Pensions Longitudinal Study (WPLS) datasets, population figures are from ONS mid-year estimates.

So to suggest, as Frank does, that there was no change in benefit receipt between 1997 and 2010 is misleading. This is the well-known statistical sin of selecting the data-points that suit your argument and ignoring those that don’t.

It gets worse. The totals for working-age benefit receipt include out-of-work benefits and benefits which are not related to employment status, notably Disability Living Allowance (DLA) and Carer’s Allowance (CA). If we exclude those claims which are only for these benefits, so that we are just focussing on people in receipt of out-of-work benefits, then the figure for May 1997 is 5.4m compared to 4.8m in May 2010.

So even with the impact of recession, out-of-work benefit receipt was lower by about 600,000. But if we compare with 2007 the absurdity of Frank’s comparison becomes glaring. In 2007, there were 4.3m people in receipt of out-of-work benefits. The number of out-of-work claims had fallen by over a million since May 1997. The sin here is that of confusing the part (out-of-work benefit receipt) with the whole (all benefits). So much for the ‘great paradox’.

We also need to take account of the reasons why people are on benefit. One of the reasons is disability. Whether people are in receipt of DLA gives us a rough proxy for the severity of their impairment: DLA is not available to people with (relatively) minor impairments. Although many DLA clients are working, severity of impairment is nonetheless clearly relevant to employment status. Similarly, if someone is caring for a disabled person, this will affect their employment chances.

If we exclude those out-of-work benefit claims which involve either DLA or CA, so that carers and the more severely disabled don’t count towards the total, then the figure for May 1997 is 4.4m, compared to 3.3m in May 2010, while in 2007, there were 2.9m claims. Thus the number of non-DLA/CA out of work claims fell by one and a half million between 1997 and 2007, and was still over a million lower in May 2010 than in 1997. The sin here is that of failing to take relevant explanatory factors into account.

There will be those who will say that despite all this, Frank still has a point about migration. He doesn’t. Confronting numbers of benefit claims with the percentage of net employment growth accounted for by migrants is meaningless. Net inward migration of working age people represents an increase in the labour supply which, as long as unemployment doesn’t rise, increases overall employment. The sin here is that of the unspecified counterfactual: what would have happened to total employment if there had been no net migration?

All that the employment of migrants shows is that prior to the recession the UK labour market was not heavily demand-constrained (if anything, it was overheated by debt-driven consumption). But who ever thought that benefit receipt, over the last decade, was driven only by demand constraints?

Rising demand from the mid-1990’s greatly reduced out of work benefit receipt, but on its own it was never going to deal with all of the factors contributing to people being out of the labour market,  such as low gains to employment, the availability of affordable childcare and disability. Labour in power did a great deal to address the first two of these factors and (despite big words) very little to address the last, with the result that long-term benefit receipt is increasingly dominated by people on DLA and CA, who account for 58% of claims running for two years or more and 65% of claims which have run for five years or more.

The Telegraph article ends with warm words for Ed Miliband:

“His speech last week, when he set out a new vision for our country based on mutual responsibility, needs to be built upon – and quickly. This will mean a historic change to Labour’s welfare policy, and he must spell out what this means in practice.”

Given Frank’s carefree approach to facts and figures, this endorsement should be treated with caution. Frank was famously invited by Tony Blair to ‘think the unthinkable’ on welfare reform. Thinking the thinkable with some regard to the evidence would be a less hubristic project, even if it meant fewer column inches in the Telegraph.

20 Responses to “Sinning with statistics: Frank Field on welfare and immigration”

  1. Declan

    Can you clarify? Incapacity benefit is included in the out-of-work benefit figures, so I don’t understand your point.

  2. Colin Thomas

    RT @leftfootfwd: Sinning with statistics: Frank Field on welfare and immigration http://t.co/FuLXKhV

  3. Stephen W

    @2.

    Yes I did read the article. It’s apparently you who didn’t you arrogant idiot.

    “Of course there is no question of the saintly Frank Field lying”

    “the number of working age claimants in May 1997 and May 2010 are indeed very close, 5.65m compared to 5,73m”

    Try actually reading what’s written rather than just leaping to assumptions that flatter your prejudices.

  4. Declan

    @1,@5
    The purpose of posts like this is to stimulate debate, not personal abuse. I don’t know who has or hasn’t read the article,or how well they’ve read it, but there is a clear misunderstanding in Stephen’s comments which I’d like to address- without insulting anyone.

    ‘So, Field’s facts are entirely correct, and relevant to the point he makes, that employment rose massively but welfare recipients did not fall, and the difference can only be accounted by the massive proportion of jobs taken by migrants.’

    On my reading, this statement is wrong on two counts. The first is the point made in the article: it is only by choosing particular dates (1997 and 2010) that it can be claimed that ‘welfare recipients did not fall’- and even then this isn’t true for out of work benefits. Between 1997 and 2007, working age employment (people, rather than jobs) grew by 2.5m. The number of out-of-work benefit claims fell by over a million. So I can see no ‘difference’ here which needs to be accounted for.

    The second is what seems to be an unstated assumption that the working age population breaks down into the employed and those on out of work benefits, leading to an inference that if a change in employment isn’t matched by a change in benefit receipt, there is an anomaly which can only be explained by migration. At least that is the best sense I can make of the statement, but correct me if I’m misreading it. A moment’s reflection shows that this can not be correct. The labour market balance is Employment =Population-(Unemployed+Economically inactive), not Employment= Population-Out of work benefit claimants. To illustrate: between 1997 and 2010, the number of (economically inactive)students increased by nearly a million and the number of working age retired by 370k, both of which increase working age economic inactivity, while other sources of inactivity reduced: for example there were 360,000 fewer people inactive because they were ‘looking after home or family’. Even were there an apparent discrepancy between employment change and benefit receipt, which there isn’t, account would need to be taken of all changes in the labour market balance to explain this. So to say that this imaginary discrepancy can ‘only’ be accounted for by migration is to give the wrong answer to a question that didn’t need to be put in the first place.

    This is not to dismiss the relevance of migration. As I said, it shows that the UK labour market was not demand-constrained over most of the last decade, which means benefit levels can’t generally be explained away as resulting from weak labour demand (this has since changed of course). That is important. And migration may have had some impact on earnings at the lower end of the distribution, potentially affecting benefit receipt, although the numerous studies on this have found only small effects- tiny compared to other factors which drive benefit receipt.

    If I’ve misunderstood anything, let me know- but on a civil footing, please.

  5. When will politicians stop taking the public for fools on immigration? | Left Foot Forward

    […] can affect benefit receipt, but not in the direct manner assumed in the musings of armchair labour market experts. Under certain circumstances, migration can affect […]

Comments are closed.