Regular readers will be aware that the cost of servicing the UK's so-called debt has been a subject of discussion here, most especially as it relates to index-linked bonds. After one recent discussion, I wrote to the Office for National Statistics, whose team in this area I have been in regular correspondence with for some time, as follows:
The blog that I write has seen rather more discussion on issues relating to accounting for national debt related issues than anywhere else I know of. The most recent discussion has focussed on the accounting for index linked bonds, which is fundamental at present to the claimed increases in the costs of government debt financing. This debate has given rise to questions on three issues, on which I hope you might be able to provide clarification. These are as follows:
1. Cost of interest on index linked bonds (ILB)
Might you please provide an example of the calculation of interest payable and capital repayable on an example ILB e.g. 20 year life, fixed coupon, issued at a premium, steady inflation barring, say a single year with an enhanced 10% inflation rate? I stress, I am in this first instance instance interested in payments due on a cash flow basis. It will help to be sure how you think these ILBs work when appraising the accounting for them. In hgve an opinion, but alternatives have been offered and the DMO example is not absolutely clear.
2. Accounting for interest on index linked bonds
- Might you explain the accounting methodologies used for ILB premiums received?
- Might you please also explain the accounting for the unwinding of this premium over the life of the ILB?
- How is interest payable accounting for? Does the accounting vary from cash paid (excepting accruals during the period until payment is made)? If so, why? Where is the other side of the ntry in the calculation of government debt if there is a variance?
- The capital to be redeemed in the case of an ILB is, of course, unknown until redemption date. How is this sum estimated by the ONS during the life of the ILB? What assumptions are used to underpin this valuation model? Do your estimates vary from that used by the DMO? If so, why?
- How is variation in the estimated redemption value accounted for by the ONS during the life of the ILB? In particular, what is the double side of the entries made and how is their conditional nature identified?
In each of these cases can you supply supporting evidence to justify the adoption of the accounting bases used?
Might you also please explain any differences in your methodology to that used by the Debt Management Office, if they exist?
Some further issues were discussed, but they are for another day.
I was told that there would be a response published on 18 July, and this morning they reminded me of this. The response is here.
I am not running at full energy today and so do not want to offer too strong an opinion when I might need more time to reflect. But it seems to me that the ONS response almost entirely ignores the second batch of questions and offers too simplistic an example to adequately answer the first.
Thoughts, please, but isn't it reasonable to think that a complex issue might be dealt with at an appropriate level of complexity? As I read it, the essence of the response is that the ONS is following rules, without actually citing the rules being followed. I think we are due something better than that on such an important issue.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Well, the key to it is “Secondly, consistent with international guidance, we record the uplift in the gilt’s principal as an interest cost each month. While the cash measure only records this cost at the point of maturity, the accruals method records this cost at the point the change in liability occurs.”
That is, they confirm that the increase in principal is booked as “interest cost”, months or years before it is paid, and whether or not it will actually be paid at all.
At the very least they need to adjust their language on “interest payments”. This is the sort of thing they say: “On an accrued basis, this month saw the third highest debt interest payment made by central government in any single month and the highest payment made in any May on record.”
https://www.ons.gov.uk/economy/governmentpublicsectorandtaxes/publicsectorfinance/bulletins/publicsectorfinances/may2022
“third highest debt interest payment made by central government in any single month”? No, no, no, no.
An accrual for an amount that will not be paid for many months or years is not an “interest payment” in any meaningful sense. It may be “payable” but it has not been “paid” so it is not a “payment”. Not yet. No cash has gone out of the door – they have just provided for a future payment.
They also have not dealt adequately with the accrual treatment for premium on issue.
Can they give a numerical example for a bond issued at a premium, like most index linked bonds are. Perhaps a real example, eg https://www.reuters.com/markets/europe/uk-sells-new-2073-index-linked-gilt-with-record-low-volume-2021-11-23/ That was £1.1 billion nominal paying 0.125% and due 2073, issued for £3.9 billion, or a yield of minus 2.3883%
I also want to see some evidence that they have included a deduction in the reported the “debt interest payment” for the unwinding premium on issue. How much was the amount reported in May 2022 reduced for that reason?
The ONS has now seen this blog and have invited more questions to be submitted
My third batch of comments to them – not included in the blog – was on language, which is precisely the issue you bring up Andrew
I will be going back with more questions – and your points are all absolutely relevant and not covered by the explanations offered.
Away now. But I will respond on Wednesday
Richard, here are my thoughts. Feel free to use and abuse in your discussions with the ONS.
First, I think it reflects well on the ONS that they have engaged with this discussion.
Second, I think their methodology is not necessarily “wrong” after all, at the end of the day (ie. the maturity of the bond) the cashflows are all accounted for accurately. However, I would ask three questions;
1. Does the way the data are presented illuminate or obscure the underlying truth?
2. Does it mesh with other data about the National Debt in a coherent way?
3. Who gives the “international guidance” that is followed and are the ONS obligated to follow it if they think it unhelpful?
The ONS explanation of why they accrue interest costs on conventional gilts on a monthly basis makes complete sense. If interest were recorded on a cash basis it would be very lumpy and the sharp rises and falls to reflect the months where large coupon payments are due would obscure the fundamental truth that, in fact, interest costs on conventional gilts are extremely stable and vary only slowly owing to the long duration of the national debt. (The two key components of the change being the change in coupon on new debt that refinances maturing debt and any extra new debt to cover the budget deficit.)
However, when I look at Fig. 1 in the paper I see a highly volatile chart with (as noted) the volatility all driven by the uplift of principal on I/L gilts. This immediately makes me ask my Q1 – is it illuminating or obscuring the truth? Should this volatile component be treated in a different way? (and no surprises here, I think it should!).
Let’s take a simple model of the national debt. It is £1,000 financed by a single 1 year I/L gilt with a real coupon of 3% (the ONS example) owned entirely by me.
Let us consider how each of us would account for the position (a) today (b) in 1 month from now…
Today is straightforward. I own an asset worth £1,000, the ONS records the size of the National Debt as £1,000.
In 1 month, assuming no changes in market yields etc., I will record the value of my portfolio as the market value of the bonds . ie. the face value of the bond uplifted by RPI plus accrued interest… which will be £1,001.70 plus £2.52
On the other hand, the ONS will record the liability as the size of the national debt – £1,001.70 ( the uplifted principal amount owing) plus any accruals that, under its methodology, will be £2.52 plus £1.65 (accrued interest and RPI uplift on principal).
In short, I think they are “double counting” the RPI uplift by accruing it in the “interest cost” data AND the “amount owing” (The size of the national debt).
So, the answer to my Q2 is – no, the way the ONS states interest costs on the National Debt is not consistent with the way they state the size of the National Debt.
My preferred approach would be for the ONS to use the mirror image of the way that an investor would view their situation. It delivers consistency and intuitively makes more sense.
Interest cost would be reported as £2.52; the change in size of the National Debt would be reported as £1.70.
Now, the size of the National Debt along with the cost of servicing that debt are important numbers and a bit more complicated than the simple model. My preferred approach would be to report:
For each gilt in issuance I would take the current face value (simple of conventional gilts, the uplifted face value for I/L gilts) multiply by the stated coupon and divide by 12 to get a monthly interest cost.
To get the size of the National Debt I would simply state its market value. I would then add footnotes that would break down the change in this market value into PRI uplift on I/L gilts, change in Market value due to changes in market yields on gilts and budget deficit.
This approach would also make dealing with discounts and premiums at gilt auctions simple.
On the issue of the APF, paragraph 6 neatly dodges the issue. Yes, interest on the portfolio and any uninvested principal (or proceeds from sales if they ever happen) is returned to the government…. But this is not reflected on the interest and National Debt data as it should be. Surely, the net interest and debt figures are the most relevant numbers.
Finally, why does this matter. Debt and the cost of servicing that debt are important political issues and will probably remain so. Therefore is it important that the ONS illuminates without bias the issues involved.
Clive
Many thanks for these points
I am now working on this issue again and will be posting my questions for the ONS in due course
Richard
Well, the latest ONS update on public sector finances for June 2022 says that “Central government debt interest payable was £19.4 billion in June 2022, which is the highest since monthly records began in April 1997; this is largely because of the effect of Retail Price Index (RPI) rises on index-linked gilts.”
So now that are talking about amounts “payable”, not amounts “paid” or “payments”. What they don’t expressly say here is that a very large part of the stated amount is an accrual for the premium that may be payable on redemption of these index-linked bonds.
The difference is clear when you look at the numbers. Table 3 has “Interest payments” in June 2022 of £19.4 billion, but Table 9 shows the cash movements of “Interest Payments” was just £5.8 billion.
It is really extraordinary that a quarter of the gilts (about £500 billion in issue) create two thirds of the accrued cost. And I’m still not convinced that the stated numbers include a deduction here for the premium paid on issue. Can they break down the £19.4 billion number between:
* interest due on conventional bonds (accrued until the next payment date)
* interest due on index-linked bonds (accrued until the next payment date)
* accrual for premium payable on redemption of index-linked bonds (which may not be due for years to come)
* accrual for unwinding of premium paid on issue of index-linked bonds (the benefit of which has already been received by the issuer)
(For the latter two, there may also be a figure for the premium or discount on issue of conventional fixed-rate bonds, if there is one.)
And are they booking the inflationary uplift in the amount due on redemption in one lump, or spreading it over the remainder of the term?
Andrew
My thanks for this
I have now got to the point of feeling well enough to work on this issue again. So first of all I have posted this, and second, will raise the points you make, with others.
Richard