Tagged: official statistics

UK Statistics Authority criticises the DWP

The UK Statistics Authority has written to Iain Duncan Smith in terms which suggest mounting concern about the abuse of statistical information.  This time it was about claims made concerning the effect of benefit cap in persuading people into work.

In the manner and form published, the statistics do not comply fully with the principles of the Code of Practice, particularly in respect of accessibility to the sources of the data, information about  the  methodology  and  quality  of  the  statistics,  and  the  suggestion  that  the  statistics were shared with the media in advance of their publication.    In March, when considering a complaint about the handling of statistics on child support, I was  told  that  senior  DWP  officials  had  reiterated  to  their  staff  the  seriousness  of  their obligations  under  the  Code  of  Practice  … The  Board  of  the  Statistics  Authority  would  welcome  further  assurance …

This has been a recurrent theme on this blog.  Many of the claims are questionable; some seem to be made up.  The complaint that the figures are being ‘shared’ with the media is also noteworthy.

The take-up statistics will be maintained

The DWP has accepted the arguments for maintaining the statistics on the take-up of means tested benefits. They cited a healthy chunk from submissions from myself and Adrian Sinfield, which is gratifying. They have not committed yet to extending them to cover the Universal Credit system, but one hopes that will follow in due course.

Official statistics and the DWP

I’ve commented more than once about the DWP’s use of stats, so it may be of interest to note a letter from the Chair of the UK Statistics Authority to the Secretary of State for Work and Pensions, occasioned by the previous statistics on Welfare to Work. Andrew Dilnot, UKSA Chair, writes:

we suggest that your Department’s approach to the publication of ad hoc statistical releases should be reviewed to ensure that all statistics which are likely to be regarded by Parliament as ‘official statistics’ are released in line with the Code of Practice for Official Statistics. We do appreciate that this is an area of uncertainty for both government departments and ourselves. The formal position is that Ministers decide whether statistical material should be treated as official statistics, not the Statistics Authority. However, the Authority has a statutory responsibility to advise Parliament on any concerns we have about the comprehensiveness of official statistics. I would be happy for Authority officials to work with DWP statisticians in finding a way forward that is mutually supportable.

I’ve also put a Freedom of Information request to the Department of Communities and Local Government to ask whether the decision of the Department not to publish statistics on Troubled Families before they were referred to and acted on has been reported as a breach of the Code of Practice.

Statistics on take-up

The DWP is running a consultation about a proposal to stop publishing statistics about the take-up of income-tested benefits. I have responded to the consultation in these terms.

1. Do you use information from Income-related Benefits: Estimates of Take-up?

Yes. My book, How social security works (Policy Press, 2011), has a discussion of issues around takeup in chapter 13, and this series is part of that discussion.

2. What would be the effect of not having this information?

This information matters. First, it provides evidence on some of the key dynamics in the process of claiming. The knowledge, experience and understanding of those involved in benefits administration and advice depends on the quality of information which is produced. For example, the practice of the administration of Pensions Credit has been based in large part on the previous learning about claiming behaviour, rather than specific findings about Pensions Credit itself.

Second, takeup is one of the principal tests of effectiveness and efficiency. For example, if we do not know how Universal Credit take-up compares to that of earlier benefits, we will not know if Universal Credit is achieving its aims.

Third, takeup is also an indicator of effective demand. The responses to the previous consultation identified the benefits of the series as including policy development, estimating benefit expenditure and future benefit counts. The series under discussion has been largely stable for some time, but that stability should not be assumed to endure in an extensively reformed system. Without information about take-up, the government will not be able to predict either the expenditure or saving to be expected from a change in rules, and any projections relating to incentives to work would be unreliable.

Fourth, benefits interact. A failure to monitor the interaction of benefits (including the new Council Tax rebates) may compromise the evaluation of welfare reform.

3. Have you any other views or comments on the proposal to discontinue the statistics?

This particular series is restricted to a narrow class of benefits. There are various theoretical explanations for problems in access and takeup. They include factors such as ignorance, lack of knowledge about personal entitlement, uncertainty about personal circumstances, complexity and stigma; they have been explained in terms of thresholds, cost-benefit calculations and attitudes to benefits. Although most of the literature on the subject conventionally focuses on income-tested benefits, none of these arguments implies that the problems are distinctive and unique to these benefits as opposed to other income-tested benefits, or that they apply to income-tested benefits and not to other types of benefit. DWP research has previously considered issues in the takeup of Attendance Allowance and Disability Allowance; it will need to consider Personal Independence Payment too. HMRC issues figures for takeup for Tax Credits, which will need to be available for Universal Credit.

The assertion that scrapping the existing statistics would save the equivalent of two full time members of staff is surprising. A repeated procedure producing consistent figures over several years from established algorithms should not require the same level of staff input as a new procedure.

4. If you are not in favour of the main proposal, which if any of the options do you prefer and why?

None of the options proposed is appropriate. At a time of major reform, it becomes more important to know what is happening, not less important. Most of the benefits mentioned are due to be downgraded or replaced and neither the same range nor a reduced range is meaningful; the range must be expanded to consider the new benefits. Given the importance of the figures, this is an argument not to discontinue the series, but greatly to extend it.

Statistics for a purpose

Part of the use of statistics, in the Coalition’s narrative of “broken Britain”, has been to show the social dysfunctions associated with benefits. The category of “out-of-work” benefits is one of the Coalition’s major innovations: it lumps together benefits related to unemployment, disability and caring in one undifferentiated and increasingly stigmatised category. The most recent release refers to people who claim benefits when they have not been born in the UK. There are some interesting and subtle findings, and the report is worth looking at. There are marked differences in different migrant groups – for example, people from Asia and the Middle East, who rights are often restrict by their terms of entry, made up a quarter of the job seekers, but more than half of those caring for people with disabilities. It emerges that more than half the people in this category are actually British or have indefinite rights to remain, and that overall foreign born migrants are nearly three times less likely to claim benefits than others. That didn’t stop the Daily Mail from reporting a fraud crackdown or the Daily Express referring to benefit tourism. Such a shame that the evidence ran in the opposite direction.

The hostile press was fed by the tone of the official statements. It is only a couple of weeks since a DWP and Ministry of Justice report revealed that 1.23 million claimants on out-of-work benefits have been convicted of criminal offences in the last five years. I find this more difficult to evaluate, because there were no comparators given in the report, and the English crime statistics have not been my constant study in recent years; I have a rough sense that crime figures are strongly skewed by age, and it might be expected that as more younger people are unemployed, and more younger people have committed offences, that more unemployed people will have committeed offences. As there have been some 8.7 million offences in the same period, the proportionate of benefit recipients with a criminal record seems if anything surprisingly close to what might be expected in the population as a whole. It appears again that the primary intention of the statistical release is to illustrate the moral deficiencies of benefit claimants.

Both these statistics have been produced for the first time. Both announcements were made by Chris Grayling MP. Looking back at the record, he was criticised last year for a tendentious and misleading release about the reassessment of claimants with disability. As Shadow Home Secretary he was told in February 2010 by the Chair of the UK Statistics Authority that his misuse of statistics was likely “to damage public trust in official statistics”. His response was that “As an opposition party, we don’t make the statistics.” Now he is no longer in opposition, he does make the statistics.

Additional note, 25th January: Today, the Chair of the UK Statistics Authority has publicly criticised the release of the statistics on foreign claimants, and the comments of ministers. “These statistics are both highly relevant to public policy and highly vulnerable to misinterpretation.”