Social Security (Claims and Payments) Amendment (No. 2) Regulations 2010

Lord Lucas Excerpts

Monday 21st June 2010 Lords Chamber

Moved By

[Lord Lucas]

To move to resolve that this House regrets that the Social Security (Claims and Payments) Amendment (No. 2) Regulations 2010 (SI 2010/870) have been based on an inadequate design which will be unable to produce reliable evidence.

Relevant documents: 17th Report, Session 2009–10, from the Merits Committee.

[Lord Lucas]

My Lords, I start by thanking my noble friend Lord Freud for turning up to respond to this Motion. The instrument that I am praying against is not of his genesis and if I get hot under the collar I hope he will accept that I am not aiming at him in any way.

The instrument first appeared before your Lordships’ Merits of Statutory Instruments Committee with a very inadequate Explanatory Memorandum. When the committee challenged them we were given more data that not only amplified what had been there before but indeed said different things. The first thing I want to say is that I very much hope that my noble friend will see the importance of making sure that Explanatory Memoranda accurately and fully reflect the intentions and details of the instruments concerned.

Secondly, although this is not in my Motion, I wanted to take the opportunity to question my noble friend on whether he really thinks that the proposals in this instrument are an appropriate use of the department’s time. It seems odd to be conducting research into what is effectively how to shift money from one government pocket into another. Here we are concerned with people on benefits. Even with the generosity of the past 13 years, benefits are still pretty marginal. If we consider the list of deductions that are generally allowed, such as mortgage payments, rent arrears, fuel and water charges, child support arrears, and so on, they are pretty important and essential to life. Yes, we allow fines to be deducted too, but that is when a court pronounces that there is a serious punishment involved. I do not think that debts to the Inland Revenue rank alongside that. Given that social security payments are basically set so that a person can meet life’s essentials, how is someone supposed to find 15 per cent of their income spare to repay the Inland Revenue? They can end up only in a worse position by having to rely on the state to a greater extent. Hence my feeling that this is just shifting money from one pocket to another.

There are other problems with this. We are not looking at large debts, but what will be the cost of recovery of these debts? Are we not looking at a system that actually costs more in its administrative functions than the money that it will recover? Is that an appropriate use of time when it comes to poor people? What is a tax debt anyway? If one is moving in and out of employment, a tax debt varies with the month. You can start off appearing to owe the Inland Revenue a good deal of money because you have not paid. You are then unemployed for six months and your allowances accumulate and you end the tax year with the Inland Revenue owing you money. It is not at all clear whether we are talking about debts that are in some way established or whether they just appear to be debts at a particular moment but may well not be debts six months later. It is an odd thing to be spending time on. This is persecuting poor people when the department ought to be trying to extract money in ways that are more efficient and have less of an impact on the very poorest.

My reason for bringing forward this Motion is the statistical inadequacy of what is proposed even with the additional explanation that I have been provided with. I know, or at least I believe I know, that the Department for Work and Pensions has a number of good statisticians—I am assured by no less than Andrew Dilnot that this is the case—but they do not appear to have been involved in the design of this research. The first thing we get is that about 5,000 tax credit customers and 5,000 self-assessment customers will be approached to volunteer to take part in the trial. Those 10,000 will then be divided into those who agree and those who do not agree. Those two groups are fundamentally different. The group which agrees will have one set of motivations and the group which does not agree will have another set. You immediately get into deep statistical water in trying to draw any conclusions when comparing two groups with different motivations. There is an attempt in the design of this experiment to compensate for that by having a third group—a control group. Presumably one could try to work out the behaviour of the first group—those who volunteer for the trial—by subtracting the behaviour of the second group from the control group and supposedly having a control for the first group.

Given the wide diversity of circumstances, we are looking at a sample that to my mind is far too small to allow that kind of second-hand approach to generating reliable results. It is a daft way of going about it when we have a perfectly good and statistically valid way of doing it, which is to double the number who agree to take part in the trial and then put only a randomised half of them through it. You would then have a matched group which will give some pretty statistically valid results. Even then the experimental design will tell you how well it works only for those who are motivated to take part in the trial. It tells you nothing about the behaviour and circumstances of those who are not motivated to take part in the trial. We are not dealing with mere mechanics; we are dealing with the effects of different methods of reclaiming money on people’s behaviour, the trouble they get into, compliance, and so forth. We are dealing with human characteristics and you cannot read those across from one group to another when you have already separated them on the basis of a fundamental characteristic such as motivation. If you wanted to find out something about the second group—those who refuse, who I suspect will be the large majority—you have to conduct a pilot that is not sample-based but area-based. The previous Government did that on several occasions. I can remember the pilots coming through the Merits Committee. A particular office would switch to a new system and would compare its results with a neighbouring, similarly placed office’s results. If you have enough other data on your claimants, you can make that reasonably statistically reliable. You could even run a randomised control through an office. But what is proposed here would not yield any useful information on the behaviour of the second group—those who refused to take part in the trial—were they to be moved to be compelled to under the new proposed system.

There is no defence in the documents that I have seen of sample size, no discussion of anticipated errors, no discussion of what are the real targets, what is expected to be achieved and what are the expected problems. There is no demonstration of the validity of the methods being set out. We have here a simple case of garbage in, garbage out. A trial on those lines will not yield any useful results. If we proceed on the basis of the results that it yields, we will have no clue whether the full, scaled-up system will work. I am comforted that the Social Security Advisory Committee, at paragraph 4.16 of its report, shares my view.

I welcome trials—they are an excellent way to make progress in the area of benefits and social security, to find out what works before one commits oneself to a whole system—but I ask the Minister to ensure that the people who really understand what are the same sort of statistical operation as drug trials have control over what goes on, make sure that what is designed is fit for purpose and ensure that the data that the trials produce can be relied on for policy formation. I beg to move.

[Baroness Thomas of Winchester]

My Lords, I thank the noble Lord, Lord Lucas, very much for the comprehensive way in which he introduced his Motion and welcome him to this most exclusive of gatherings in London; the tiny number of us in this House who speak on DWP orders and the even tinier number who table regret Motions to them.

Before going any further, we have discovered from the response given by my right honourable friend Steve Webb to a letter from the chairman of the Merits Committee last Session that the new Government may decide not to go ahead with this pilot, but I suppose that we must carry on as though they will go ahead at this stage, unless the Minister wants to interrupt at this point to say that they will not. There is no interruption, so I shall carry on.

First, I am glad that the Government have listened to the Social Security Advisory Committee's recommendation, which urged them to bring the list of benefits from which HMRC debt recovery repayments can be made into line with the current list of benefits from which priority debts can be deducted. In other words, any deductions would be from means-tested and not contribution-based benefits. I take the point made by the noble Lord, Lord Lucas, about that being money from the poorest people in the country, whichever benefit it is from.

I am also glad that the SSAC highlighted the vital point that claimants understand the voluntary nature of the trial and the impact on their income of signing up for the trial. A lot of people will not know what “voluntary” means—although they may pretend that they do. The Government have responded positively to that point, saying that they will share copies of the letters sent to claimants with the committee, and that they will ensure that the letters, and contact centre staff, will direct the claimant, or customer, to the availability of independent advice.

The SSAC understandably believes that there is scope for confusion among claimants who receive letters from HMRC. Just seeing that letterhead is likely to lead to many claimants into thinking they are being hounded for recovery of their tax credit overpayments, whatever the letter actually states. If claimants then telephone HMRC, is the Minister satisfied that they will be told in every case that they do not have to repay their debts by having them deducted from their benefits? Will they be told in all cases that they might want to take independent advice, which might lead to some of their debt being overturned or even written off?

Before leaving the subject of letters from HMRC, perhaps the Minister can tell us why it does not provide an explanation of a tax credit overpayment to claimants in all cases; that seems not to be provided as a matter of routine. I cannot think why not. I know that the Minister does not speak for HMRC, but as this SI is a joint project between the two departments, I make no apology for asking him that on this occasion.

We must remember that the Government did not agree with the SSAC that repayment should be at a lower level than they are proposing in this trial, which is three times the normal amount per week. As the noble Lord, Lord Lucas, pointed out, benefits are not generous, and there are rumours that they will not be uprated as usual—we shall hear more in tomorrow's Budget. If claimants in debt have no other source of income except their weekly benefits, many of them will suffer severe hardship if they are being encouraged to repay their tax credit overpayment from those pretty meagre benefits. I may have got the wrong end of the stick here, but I find it horrifying that the SSAC understands that HMRC staff may suggest that debtors apply for a commercial loan to pay off debts—presumably they would never suggest a commercial loan to pay off a tax credit overpayment.

Turning to the design of the pilot, and first, to the aim of the pilot, this appears to be twofold: to find out if there is a demand from claimants for this method of repayment—that sounds almost like an oxymoron—and to see if the scheme is cost-effective. There is more information in the answers to questions from the ever-vigilant Merits Committee about how the scheme will operate. As the noble Lord, Lord Lucas, said, there are to be three groups for evaluation purposes. The first group will be those who volunteer to take part in the trial. The explanation goes:

“The evaluation will look at how much was recovered, the spread of weekly deduction rates, how long a person is in receipt of a relevant benefit and what could have been recovered had they been taking part in the trial for the whole two years taking account of expected levels of movement on and off benefit”.

That sounds fiendishly complicated with so many permutations that you wonder whether any comparisons will really make sense. The next two groups are about those who declined to take part in the trial who will form the control group. First, they have to agree to respond to a letter asking them to take part. I wonder how many will be keen to do that. Not very many, I imagine. Therefore, a key question is: how small does the sample size have to be to produce robust data? I note that the Government say:

“We accept that the smaller the sample, the less able we will be to draw definitive conclusions."

I also note what the Government say about those who go off benefit during the pilot period; namely, that there is no minimum number of weeks for which deductions have to be made in order to regard the participant and repayments made as significant for evaluation purposes. I would have thought that was a rather significant fact.

Perhaps the most encouraging sentence in the whole explanation is:

“In addition the performance of new joined up operational processes between HMRC and DWP will be assessed”.

Is the Minister really confident that this trial will produce a reliable result, in view of all the problems that the SSAC and others have pointed out?

--- Later in debate ---

[Lord Freud]

I will not rise to that.

My noble friend Lord Kirkwood asked when the trial was due to start. It will start next month, and the first letters will go out then. We do not intend to cancel this trial; we will go ahead with it, although we will extend it only if it is successful.

The noble Lord, Lord McKenzie, asked whether other methods of repayment would be available. Clearly, if the customer signs up for the trial, no other method will be used. If the customer does not sign up, the repayment methods could be lump sum or instalments by direct debit or standing order. He also asked whether the new Government stand by the decision to deduct at three times the 5 per cent rate. Yes, we stand by that, but it is the maximum rate and participants can choose a lower rate. Finally, he probed the question of other ways of applying sanctions. This is, of course, not a sanction—it is a repayment of a debt—although I can tell him that we are exploring non-financial sanctions.

I commend the principle of the trial as a convenient alternative repayment method for those who wish to use it and as an example of joint working. The design of the trial is adequate, but I accept the criticisms of the Merits Committee and of my noble friend Lord Lucas. We need to get this right in the future. We need to make sure that all our regulatory changes meet the standard expected by the Merits Committee and that we provide all the necessary supporting information in good time.

[Lord Lucas]

My Lords, I am extremely grateful to my noble friend for that comprehensive and helpful reply. I shall not trade blows with him on matters to do with welfare, as I am merely an extremely junior acolyte at his feet. This has been a fascinating debate for me and I shall feel tempted, now that we have a Minister in this House who is so much at the forefront of welfare reform, to sit on the Benches and, at least, to listen. However, I am quite happy to trade blows with his statisticians.

The Minister has made me understand one thing that I did not understand before. The trial is clearly voluntary, but is he saying that the rollout will also be voluntary, even if the trial is a success? That certainly would remove a lot of my worries. However, the effort being expended to create a control group and to look at what is happening in the second, probably larger group of those who will not take part in the trial rather suggests to me that there was an intention—noble Lords opposite may know whether this is the case—that this should be a compulsory way of reclaiming HMRC debts.

If you want to understand how to operate this effectively with people who volunteer, you do a randomised control trial just with those people. You go on until you have a couple of thousand volunteers and you assign them randomly—half you monitor under the current system and half you monitor under the new system. You then have a conventional and statistically robust way of comparing behaviour. It might be fun and informative, and it might have a peripheral virtue to try to understand why the people who have not volunteered have not done so and to try to find out how to encourage them to do so in future, but you certainly would not bother with this functionless control group that sits as an appendage at the bottom.

As I say, there is a simple and statistically robust way of dealing with this if the trial is for a voluntary system, which would not have all the characteristics of the trial that has been put in front of us. If the Minister is prepared to set up a meeting with his statisticians, I should enjoy it very much, because I do not think that they have come up to the mark on this occasion if, as I say, this is a trial for a voluntary system. If I am wrong and this was designed as a trial for a compulsory system, I come back to my old criticism that it does not function as that. You do not get enough information on the likely behaviour of the people who have declined to take part in the trial to be able to predict how they would react if they were compelled to take part.

However, I cannot be churlish when I have received such a good reply from my noble friend, particularly if he sets me up with a tea date with his statisticians—and how could any statistician refuse tea in the Lords? Whatever the circumstances, I happily beg leave to withdraw the Motion.

Motion withdrawn.