DWP responses to my suggestions for changing the assessment of earnings.

It’s some time since I posted to this blog, not because nothing of interest, or worthy of comment, has been happening, but simply because, like many others, I have been busy.  In that, I am lucky; far too many people have been without work, and much else over the last months.

I’ve kicked myself into making this post, because my suggestions about changing the assessment method of earnings for Universal Credit got a little attention this week.

I gave written and oral evidence to two recent inquiries, the Lords  Economic Affairs Committee inquiry, The economics of Universal Credit and the Commons Work and Pensions Committee inquiry, Universal Credit: the wait for a first payment.

My suggestions are summarised in a post on this blog here and an explanatory video on YouTube here.

Both inquiries delved deeply and widely into a number of issues and took evidence from a  large number of sources.

The Work and Pensions Committee report recommended that the department should assess my proposals.

Recommendation 32

We recommend that the Department assesses a model in which it estimates people’s daily pay rate from data it already receives from HMRC, as suggested by Ferret Information Systems, and in which it would make payments from day one of the claim with reconciliation in month two. We request that the Department share information on what data that it receives from HMRC, including whether it knows how frequently people are paid, and makes clear what parts of its system are and can be automated.

 In its response to the report, the DWP replied:

The proposed system described by the Committee cannot easily be operationalised because, whilst you can conceptualise it for an individual case with regular earnings, you cannot reliably run it where earnings fluctuate (such as in zero hour contracts) or where there is more than one set of earnings data (such as in couple households where both maybe working, or individuals with more than one job). In these instances, the reconciliation would be challenging and likely to result in over and underpayments each month as earnings data is verified. This would lead to financial uncertainty and a poor customer service.

HM Revenue and Customs sends information to the Department relating to UC claimants who are employed, this includes: gross taxable earnings less tax, National Insurance and pension contributions, pay dates and frequency of payments. This information is combined with any self-reported earnings and income to calculate awards of UC at the end of claimants’ assessment period.

In a unique joint session of the House of Lords Economic Affairs Committee and the House of Commons Work and Pensions Committee on Tuesday 9 March 2021, Will Quince, Minister for Welfare Delivery and Neil Couling, Senior Responsible Owner Universal Credit at DWP, gave evidence relating to the  recent inquiries by the committees.

During the session they made reference to my evidence and suggestions.  I’ve put up some extracts from the Parliament TV record of the session here 

 While their comments are kind in many respects, I think that some reply is justified to a few of the comments that were made during the session and in the earlier response.

The response first:

“you cannot reliably run it where earnings fluctuate (such as in zero hour contracts)”

It is true that this may make the following reconciliation more variable but that is still unlikely to produce more variation in payment than the current system and, if estimates of earnings are realistic, considerably less.

“you cannot reliably run it …where there is more than one set of earnings data (such as in couple households where both maybe working, or individuals with more than one job).”

Wrong; I have demonstrated in my proof of concept that it deals with these situations as easily as it can deal with one single regular wage.

“…the reconciliation would be challenging and likely to result in over and underpayments each month as earnings data is verified. This would lead to financial uncertainty and a poor customer service”

I recognised in my evidence that, in cases of irregular earnings, there would be a need for reconciliation of some over or under payments. These would though be much lower than the structural variations which exist in the current system, which the department praise for producing excellent customer service.

From the evidence session this week:

“Hypothetically it could work and would, work …”

The minister said

“Hypothetically it could work and would work but it would rely on very accurate recording of start and end dates.. as well as people working regularly within a period.”

The first point is true, but applies equally to the assessment of tax and national insurance, by HMRC, which is a part of the data, from HMRC, used in the current assessment method. The second point is not true as my suggestion applies accurately to those working in very variable situations.

“What if they have a partner with a different working pattern and different daily rates”

My suggestion, as is shown in the video explaining it, handles this type of situation, and the other examples mentioned, accurately and without the exaggerated variability that this situation can produce in the current situation.

Neil Couling described the current system as “simple sets of rules, that apply to all cases” and his determination to make sure that layer upon layer of changes to rules would not complicate the system, made him resist these kinds of changes.

The 68 amending sets of regulations, that have Universal Credit in their titles, make the success of this approach somewhat uncertain. He said that the Treasury would not allow the introduction of a ‘simplicity rule’ that ministers would have had to satisfy before bringing in a change’. I wonder what Universal Credit would look like today, if that had succeeded.

“It is very clever and very attractive, but it just cannot be made to work”

Kind words from Neil Couling, to end with. My closing comment is simply “I don’t understand why not”.  It won’t be as simple as with my spreadsheet demonstrating it, but the actual calculation side of Universal Credit is a tiny bit of the system and that is the piece that would be changed.  The data is already there and has to be used anyway.

How about a trial?

Here’s a challenge, give me an anonymised set of data – small enough for my feeble PC to work with but big enough to count – and I’ll run a model of the process.  Then there’ll be some evidence to look at.