The obsession with fraud is not new

A little bit of myth-making, this time about social security fraud.  Zoe Williams writes in today’s Guardian:

Even in the darkest days of me-first Thatcherism, the social security conversation hinged on whether or not the dole was enough to provide a decent life.  … the question of fraud rarely came up.

In fact, the obsession with fraud pre-dates Thatcherism.  Golding and Middleton’s book, Images of welfare, published in 1982, attributed the start of the moral panic about fraud to reporting of the Deevy case in 1976, but it started some time before that.  I’d date it from the publication of Robin Page’s exposé in the Spectator on 6 September 1969.  The article was syndicated in the News of the World (two weeks running) and questions were raised in Parliament.  In 1971, Keith Joseph set up the Fisher Committee on the abuse of benefits, which reported in 1973.   When the Thatcher government came into power, one of the first steps in this field, taken early in 1980, was massively to inflate the fraud figures.  (Reg Prentice explained to Parliament that higher figures were used by ‘large commercial organisations’ but said there was no reason to do any work to check that assumption.)

The obsession with fraud has been poisoning the system for decades, and there is no evidence that anti-fraud measures have done anything to improve the situation – the auditors haven’t fully approved the DWP accounts for years.  There is an alternative. The estimates for fraud and error in the State Pension suggest overpayments of 0.1%; the estimates for Pension Credit, a benefit which goes to more or less the same group of people, comes to 5.6% – more than fifty times as much.  If the government was serious about reducing fraud and error, they should look at systems which deliver benefits accurately and efficiently without it.

The Irish Department of Social Protection is clamping down on people with false beards

There are times when the surreal breaks through the curtain to engulf our dull grey world, and a note from Ireland gives me the sense that another such moment is upon us.  A new fraud initiative based on facial identification promises to cut off the presumably lucrative trade in disguises and false beards.  According to the Irish Independent, the minister for Social Protection has reassured an apprehensive public that “Welfare cheats using make-up and fake beards to get benefits won’t beat the ID software.”  That must be a relief.  I do not know how many cases there have been of theatrical imposture undermining the integrity of Irish benefits, but given the expense of introducing such a system I am sure the threat must have been substantial.

The Tax Credits fiasco

I’ve spent much of the afternoon listening to the proceedings of the Work and Pensions Committee concerning Concentrix. HMRC had identified 1.5 million Tax Credit cases where they had concerns; they engaged Concentrix, a private firm, to process cases.  Concentrix was reported to have sent out a million letters fishing for information, challenging for example whether they were not living with an undeclared partner; the firm’s representative told the  Committee that they had sent out 324,000, though over a shorter period.  People who did not reply to the letters had their benefits stopped; 90-95% of those who asked for reconsideration had the decision overturned (that was Concentrix’s estimate – HMRC gave a lower figure, of 73%).  HMRC had terminated Concentrix’s contract, but they seemed much more concerned about the collapse of the phone line in August than about the huge number of wrong decisions that their policy had generated. Frank Field MP, the Chair of the Committee, told the Independent:

The Committee was astonished by the extraordinary evidence we heard. From Concentrix we saw a company desperately out of their depth and unable to deliver on the contract awarded to them by HMRC. From senior HMRC officials we saw a palpable disregard for the human implications of this gross failure of public service. From the tax credit claimants we saw dignity in the face of appalling and traumatic experiences.



The statistics for fraud and error wobble around

As part of my work on the new social security book, which will be out in February, I’ve been updating some of the figures I refer to.  I’ve just been revising figures related to fraud and error in the benefit system, and I’ve been struck by the extent of variation in the DWP estimates by comparison with previous figures.  Overpayments of JSA are said to have increased from 3.9% to 5.0%, both because of higher fraud and higher official error.  Overpayments of Pensions Credit came down from 5.9% overpaid to 4.6%, but have now apparently gone back up to 5.6%, even though claimant error seems to have fallen.  Over the last couple of years the figures for Housing Benefit have reapportioned blame from claimant error to fraud.   This is an uncertain area, and most of this can probably be put down to methodology rather than any underlying trends.

The HMRC figures, which refer to a previous period, avoid problems by a simple strategem: they don’t admit to making any mistakes themselves.  (Table 5 in their figures puts the damage at no more than £10m out of nearly £30bn).


Fraud and error in Universal Credit

Universal Credit was supposed, through some process I can’t claim to understand, to reduce fraud and error in the benefit system.  According to the Business Case, there were supposed to be reduced overpayments worth £14 billion over twelve years, or £1.16 billion per annum.    I commented in 2012, and again last December, that I found this claim implausible: Universal Credit doesn’t do anything about most of the circumstances which lead to fraud and error, but it does add layers of complexity to make confusion and mistakes more likely.  Now we have the first figures from the DWP on fraud and error in Universal Credit.  Universal Credit was overpaid by 7.3%, compared to 5.0% of JSA (the nearest comparable benefit) and 5.2% of Housing Benefit.  It was underpaid by 2.6%, while JSA was underpaid by 0.8%. The system makes more mistakes in both directions.  The DWP points to the complexity of managing some claims – but that, of course, is the point.

The DWP summary pleads that the difference between UC and JSA is ‘not statistically significant”, and that in any case UC and JSA are not directly equivalent.  True enough, but Universal Credit was supposed to cut fraud and error by somewhere between a third and a half of all overpayments.   If UC was supposed to save a third but is actually costing nearly fifty per cent more, the level of fraud and error is running at more than double what was expected.  If it was supposed to save half, it’s three times what it should be.

Fraud and error statistics: the biggest problems relate to employment, defining households and capital

The statistics for fraud and error in the benefits system were released last month, and I’ve just caught up with them.  The summary emphasises that ‘fraud is the largest cause of overpayments’, but that’s only because error is split into two categories – the official estimate is that out of £3 bn overpaid, £1.3 bn is down to fraud and £1.7 bn is down to error.

The accompanying tables raise some interesting questions.   The reasons for overpayment that feature most strongly are employment and earnings, household composition and living together, and capital (Table 3).   The main reasons for underpayments  are employment and earnings, household composition and confusion about the relationship with other benefits.  This shouldn’t be any great surprise.  Despite assumptions to the contrary, people don’t necessarily know whether they’re working, whether they’re in a relationship,  whether they’re disabled and so on, and the more sensitive the system is to these issues, the more likely there are to be mistakes.

Drilling down a little further, the benefit where there is most confusion about employment and earnings is Housing Benefit for people of working age – this category alone accounts for £315m of overpayments and £131m of underpayments.  Wherever problems emerge repeatedly at particular points, there’s an argument to review the process, and this is no exception.  Unfortunately, with the transition to Universal Credit there’s not much likelihood that this will get any attention at all, and we should be wary of the assumption that Universal Credit will clear all this up.

Errors in Housing Benefit payment: another failure of personalisation

The Public Accounts Committee has been highly critical of the DWP’s failure to deal with ‘fraud and error’ in Housing Benefit, estimated at £1.4 billion last year, 5.8% of total expenditure.  The detailed figures came from a report by the Comptroller and Auditor General last October.  Despite the tag, “fraud and error”, the biggest part of this is error:  £900m is due to mistakes by claimants, £150m mistakes by officials and £340m in fraud.  These figures don’t include the other side of error, which is underpayments:  £370m of the benefit is underpaid, of which £290m is due to mistakes by claimants.

The biggest single source of the mistakes is the mis-statement of earnings, which accounts for a whopping £637m (46%) of the overpayments.  This is what you get with  a ‘flexible’ labour market: people don’t know what they’re being paid or when.  The next main sources of the mistakes, 16% and 11% respectively,  concern people living together – confusion about what makes a household – and residency, where people are no longer living in the property.  In some cases, those run together.  People who are forming relationships often occupy two properties for a period.  They can’t give a definite answer about where they live and whether they are now a household.

We ought to be asking whether it makes sense to try to run a benefit scheme on this basis.   The basic problem lies in the belief that it’s possible to ‘personalise’ benefits and to adjust them as individual circumstances change.  It has never worked properly before, and there’s little reason to suppose it will work  in the future.

Prosecuting fraud (again)

I contributed yesterday to Radio Scotland’s ‘Call Kaye’ programme, discussing the proposition that social security fraud needs to be prosecuted more vigourously.   The prominence of social security fraud in reports has always been troubling.  The constant harping on fraud, over more than forty years, has poisoned relationships between claimants and benefits agencies.  The level of fraud has been greatly exaggerated; the focus on fraud in benefits, estimated to be 5.5%  of fraud against the government, is disproportionate; reports about individual frauds are usually inflated, because they discount the money that people should have received instead; sentences are already harsh.   Rather to my surprise, most callers took a similar line.

It’s fair enough to say that where there is deliberate fraud or organised crime, it’s anomalous to treat it as a distinct offence under the social security legislation – there is not much difference in principle between defrauding social security and defrauding a bank.   At the same time, what people mean by ‘social security fraud’ is not usually ‘fraud’ in that sense.  People receive money they’re not entitled to for all sorts of reasons, most usually when their circumstances change and they don’t inform the benefits administration.   The three largest categories are when they have an occupational pension which is not declared; when they are sleeping with a partner they haven’t declared; and when people get some work and don’t declare it.  Those are not quite in the same class as counterfeiting identity papers or hacking the payments system.

The problem with occupational pensions is an example of a more general issue; people aren’t always clear about what has to be declared and what doesn’t.  When we had insurance based benefits, other sources of income didn’t matter.  The second problem, ‘living together’, is one I’ve referred to several times; the questions are intrusive, people aren’t open about their relationships, and while sometimes people lie, others couldn’t describe their relationships in the terms that are required.

Working while claiming seems to excite the greatest passion and condemnation – which is perhaps surprising, because people who are working are doing exactly what successive governments say they want people to do.   If you want to get yourself job-ready, to show willing to work, or to move up the ladder into secure employment, working whenever the opportunity presents itself is surely the way to do it.   However, the flexible job market doesn’t give security, adequate wages or  even certainty about whether or not you’ll be paid – and people in that position are ‘fiddling’.  There are ways to organise benefits that allow for some casual work – for example, providing benefits over six months (the old rule for Family Income Supplement), taking income over a six week or three month period, and of course providing universal benefits like Child Benefit that aren’t touched by fluctuating incomes.  What we’re about to get instead in Universal Credit is a system that is going to bounce the level of benefits up and down like a yo-yo, while demanding that claimants bear the cost of any mistake, regardless of who has made it.

More statistics, more scare-mongering

The Daily Telegraph  reports that “Fraudulent and wrong benefit claims hit £3.5bn record”.  There are two obvious points to make about that claim.  The first is that this figure is based on a combination of fraud and error, and the largest part of the figure is error.  The estimate for fraud is £1.2 billion out of £3.4 billion.  The second point is that this is a very long way from being a ‘record’, if by that they mean a new high; the 1997 Green Paper on fraud put the figure (ludicrously) over £7 billion.

Reducing fraud and error?

In Parliament on Tuesday, Iain Duncan Smith claimed twice that the Universal Credit system would reduce fraud and error by up to £2.2 billion. He explains:

“Many people are receiving overpayments or underpayments when they should be receiving the correct amount. Too often with tax credits, people are chased at the end of the year, without their realising that they had received the wrong money in the first place. Universal credit will be kinder in the sense that it will be adjusted each month. It will help us save huge sums — some studies state £2.2 billion per year.”

That sum ought to be be compared to current estimates. Currently the total losses through error and fraud in DWP benefits are estimated at £3.4 billion, while for Tax Credits and Child Benefit the estimate is £2.3 billion. £460m of the first figure is down to Pension Credit, which is not part of the Universal Credit system, so the total comes to £5.2 billion – nearly half of that is down to mistakes made by claimants. But there are also mistakes made in favour of the government – estimated at £1.4 billion. That means that the difference between overpayment and underpayment is £4 billion, and Duncan Smith is claiming that well over half of that is going to be saved in the new system.

I’ve expressed doubts before as to whether the reformed system will save anything at all. Table 6.1 of the newly published statistics on fraud and error gives a long series of categories where errors have been generated: they include, among others

  • Capital
  • Conditions of Entitlement
  • Household Composition
  • Housing Costs
  • Passporting
  • Non-Dependant Deductions
  • Earnings/Employment
  • Income Other Benefits
  • Income – Occ & Pers Pensions
  • Income – Other
  • Control activities not being carried out
  • Living Together
  • Labour Market Issues

Three factors together – earnings and employment, income from occupational and personal pensions and living together – make up 44% of all the losses through fraud and error. However, most of the list will apply to Universal Credit with as much force as it does to benefits now.