THURSDAY 16 SEPTEMBER 1999

DR HELEN BARNES, DR GILLIAN PAULL and PROFESSOR IAN WALKER

  420. Has the Minister had the benefit of having a chance to look at that? Is this something that it would be fair to cross-examine the Minister on later this morning?
  (Professor Walker) Some of the issues of principle have been exposed to the Minister but not the precise numbers.

  421. You were making the more general point earlier that really there is not enough research and you are trying to fill that gap and that is very welcome but researchers always say that, do they not? If you were the Government and you are where you are and you are looking at situations where perhaps we are taking decisions based on inadequate amounts of information, how long do you think it would be realistic to wait to get appropriate levels of information before you would be confident that the policy makers are able to get a firm grasp of what the implications of the various reforms that they are contemplating are?
  (Professor Walker) It is very difficult to say because evidence comes from two basic sources. Firstly there is experimental evidence where you change something and then you wait to see what impact that has had on a group of those subject to the change relative to the people who have not been subject to the change. That has been done in the United States but unfortunately only over the last three years and still there is no systematic analysis that has taken place across those states.

  422. You understand that there is a pressure on us. We have to see people every Friday and Saturday and people are saying this system is broken, it needs to be fixed very quickly, and if we say "do not worry, the researchers are on the job", I do not know that that would leave them any happier.
  (Professor Walker) I think that is why what I am asking for is for the DSS to consider implementing this in a way that enables us to fine-tune it in the fullness of time.

  423. Does that mean pilot projects or does it actually mean further research work in the universities? What does it actually mean? How long does it take and how much does it cost?
  (Professor Walker) In principle, it should not cost very much because if you do this as part of the implementation of the reform, then it is simply a matter of analysing the administrative data that is then forthcoming.

  424. Do I understand you to be saying that in spite of the work that you have just been unveiling to us this morning, you are willing to let this White Paper reform unfold in an evolutionary way and as long as you monitor the way it is going during the transition and all the rest of it and get the research work done at the same time in parallel with that change, you would be content to let it roll as it stands?
  (Professor Walker) Well, this is what we should have done four or five years ago, but did not. The prospect of making fundamental improvements to the way in which the system works is not just a case of administrative simplicity because actually we have US evidence to suggest that actually simplicity does not necessarily promote compliance. If simplicity is bought at a cost of generating liabilities that are less related to the needs of the qualifying children, then the US evidence suggests that compliance will fall, so if the reform is implemented in a way that makes fathers think that the resulting liabilities, although smaller, are somehow less related to the needs of their children, then compliance could actually fall, so there is a great deal that we do not know because the only evidence that we do have is non-experimental evidence. Our only option is to look at what we can learn from observing people's behaviour and this evidence is, as I said, only a censored view of the whole population.

  425. Would it be sensible to base legitimate, full-blown research on data that comes from the CSA because we have been hearing earlier that nobody believes anything they say anyway?
  (Professor Walker) Sure, the CSA is notorious for its measurement error.

  426. How are they going to help you do robust scientific work?
  (Professor Walker) Statistical methods can be used to deal with data that is not as clean as experiments in natural science would generate. Statisticians do this all the time; they are dealing with data that is contaminated by errors and you would not necessarily want to design a system to generate data like that, but you can deal with it once you have it. You might, for example, argue that, on average, the measurement error is zero. That might be one way of doing things and then you can say, "Let's compare this group of people who, say, have wage-withholding applied to them and see, on average, how they differ from people who have not had wage-withholding applied to them".

  427. I believe you! I am not sure about it, but you are the scientist and you should know.

  428. On the point you are making about the effect on working families and the income support disregard based on the American experience, can I ask you if you have factored into that contribution the position in relation to how the caring parent's income is treated and is that consistent throughout the states or different between the states? I would have thought that would be the key factor in relation to work incentive as part of the overall package.
  (Professor Walker) The work incentive numbers that I mentioned before are simulations based on British data. The reason why the American evidence is really hard to interpret is that although there have been systematic differences across states in the way in which the reforms have been implemented which ought to be helpful, but you cannot actually isolate one aspect of those reforms in isolation. So, for example, the AFDC disregard in the child support system of the United States was changed in different ways in different states, but lots of other things changed as well in different ways across different states, so it is going to be a complex problem to unravel what the implications any one feature of the child support system is likely to have because all sorts of features of the child support system and the welfare system have been changing at the same time. The evidence we have in the UK is non-experimental—it does not extend to looking at what has happened to one group of people compared to another group of people; it just says, "Let's look at the people and see how they are different in observable ways and try to relate those differences to how they behave". That is the background to the table you have on labour supply effects, basically just looking at how people differ in terms of their incomes and wage rates and so on and trying to relate that to their behaviour and then using that statistical model to predict what would happen if we changed their income and wages and so on which the reform will ultimately do. We can see that the result here is that actually the income support disregard does not seem to be damaging to labour supply; it is compensated by other aspects of the reform. The worry then about the child support disregard for income support is that it actually may not do anything to promote compliance.

  429. This is all interesting stuff, so I want to slip into the Chair's wet towel for a few minutes and see if we can get our heads round some of this stuff. Perhaps I can start with Professor Walker's table here. The changes look relatively small. We know that there are real difficulties with models on incentives and balance of income, substitution effects, et cetera. Are these changes significant, for instance, say, for the non-working households, the 53 per cent pre-reform with family credit to the 50 per cent pre-reform on working families tax credit?
  (Professor Walker) Yes, these are based upon behavioural estimates which are statistically significant, so switching from, say, 52 per cent non-working to 48 per cent non-working is actually quite a large change. Taking close to 10 per cent of lone mothers who currently do not work and turning them into workers, I think, is quite a substantial achievement.

  430. You say that the income support disregard is not damaging to incentives at the £10 level. Is it a fairly simple job to run these simulations again, say, at £15 which is the level which has been proposed by a number of organisations?
  (Professor Walker) Absolutely.

  431. I think it would be helpful if we could have those figures.
  (Professor Walker) Yes, no problem[6].

  432. If I could turn to Dr Paull's work, I assume that this is based on no behavioural changes at all?
  (Dr Paull) Absolutely.

  433. What you are saying to us is that actually the reforms themselves have very little impact on child poverty overall. The major impact is the move from family credit to WFTC which, if I have understood your figures in terms of the proportion of children in families with care, falls from 43 per cent to 37 per cent as a result of that change, assuming no change in compliance.
  (Dr Paull) The move from family credit to working families tax credit has very little impact on the proportion of children in caring mother families in poverty and it remains at around 43 per cent. The change in the reform, assuming working families tax credit from the current child support system to the new child support system, assuming no change in compliance also, has very little impact on the proportion of children in poverty and it goes down from 42.5 to 41 per cent and if compliance increases to 80 per cent, that is still 40 per cent of all children. Where it did have a larger impact is if we just look at caring mothers who are on income support at present, which is just 55 per cent of the group, then the proportion of children in poverty falls from 64 per cent to around 60 per cent if we have 80 per cent compliance under the new system, so it is still not huge numbers.

  434. You are sceptical about increasing compliance anyway. In the earlier submission you said that is because of the machinery problems of increasing compliance but presumably it is also to do with the ability to pay of non-resident parents?
  (Dr Paull) Yes. We do not know what direction compliance will go in. What our work stresses is that even if we do get very large increases in compliance it is still not going to have a very large impact.

  435. One last request from me and you do not even have to answer except to say "yes" or "no". Is it possible to run these calculations on the basis of the £15 disregard as against the £10 disregard to let us see what the effect is of that?
  (Dr Paull) Absolutely.

  436. I found difficulty in understanding the presentation and I have got a plea to institutes like yourselves to use ordinary language. You say in eight: "The decision to make all non-resident parents pay a minimum amount of child support is supported by two main rationales: the symbolic honouring of the child support commitment, and the creation of a normative expectation of payment." I presume what you mean by that is that it is right to make fathers pay in discharge of their responsibilities. What I really want to ask is why did you come up with this view, which I think is quite important, that non-resident parents with housing costs and those whose former partners have high incomes will be worse off?
  (Dr Paull) That comes from the change in the formula where under the current system those who have higher housing costs have lower child support liability. Under the new system those housing costs will not be included, therefore their child support liability will be higher assuming that everything else is constant and similarly because the other incomes are now included. I do not think we have looked at this directly because other changes in the reforms could mean that certain individuals within those categories could be better off or worse off.

  437. I think it is quite important because if one could have any direct research on the ones with high incomes because they will be worse off, that is the absent fathers, I think that is quite an important point.
  (Dr Barnes) They are a very small minority of people as the White Paper pointed out on page 49.

  438. They talk about 6,000 people, do they not? Is that the 6,000 figure or is that another figure?
  (Dr Barnes) Those are the ones with shared care. Yes, it is around 6,000.

  439. If you have got anything clear on that I would be interested to have that because I think it is an important point. It is a point that has been put to us and it is likely to cause controversy.
  (Dr Barnes) Could I just come back on that. That is something that we raised at the end of Trial and error, something that although it only affects a minority of people is a really bad constituency case. It is the sort of thing that makes really bad headlines although it only affects a minority of people and it could lead to a lot of political pressure. There are ways of dealing with it like the Australian system where you actually take account of the parent's income above a threshold and what you can actually do is set that threshold at a level where you only bring in those cases where income is very high, so you are not muddying the system but you are able to close off that gateway. I have set that out in my submission.