The ONS claim that government interest costs were £8.1 billion in December looks to be seriously overstated

Posted on

According to the Office for National Statistics release on the December 2021 government finances, published yesterday, this happened:

That's an extraordinary increase in interest payments for a month. It makes it worth looking at.

This is the composition of UK national debt at present according to the Debt Management Office:

Market prices were above issue price:

The premium is 15.3% on conventional gilts in September 2021 and 18.4% in March 2021.

To summarise the split was as follows:

And the rates of return were as follows:

And just for the record, the gilt holdings were as follows:

33% of all gilts are owned by the Bank of England. None of these are index-linked.

To understand index-linked bonds we need to know changes in the retail price index to which they still appear to be linked according to top the data I can find. Over the last thirty years or so RPI moved like this, based on ONS data up to the end of 2021:

To be precise, this index was measuring 1.25% annual change in December 2020 and 7.16% in December 2021, an increase of 5.91%.

So, let's summarise this data. Using net debt in issue (i.e. allowing for bonds owned by the Debt Management Office itself:

  • There was technically £2,023bn of debt in issue
  • Of this £480bn was index-linked
  • At issue price £668 billion of government debt was owned by the Bank of England - a figure that seems very low given that £875 billion has been spent on this debt but which roughly accords with other data from the ONS, so I will accept it
  • This leaves £885 billion of conventional debt actually in issue.

Now let's estimate interest owing.

  • Most index-linked debt carries little or almost no notional interest - anything between 0.125% and 0.5%. I will use 0.3%.
  • The average noted yield on conventional bonds across the portfolio was 0.94% according to the Debt Management Office, which was up by about 0.2% on a year ago. However, this is calculated on market value. That means on nominal value rates need to be inflated to allow for this.

So, using a simple calculation and assuming little real debt issue in the year - which is true because of quantitative easing, then:

  • In December 2020 conventional gilt interest cost was based on net debt of around £900 billion at about 0.75%, divided by 12, of course, or less than £0.56 billion. If the Bank of England owned debt was included it would be around £0.9 billion. Allowing for rates on nominal rather market value this can be rounded up toward £1.1 billion in cash cost.
  • Index-linked bonds in December 2020 would have yielded maybe 0.3% plus RPI or about 1.55%, meaning the costs was maybe £0.6 billion

It's hard to see where the figure of £2.7 billion comes from on this basis.

In December 2021 the same figures would be:

  • Conventional gilts, £885 billion at 0.95%, or £0.84 billion. Grossing this up to allow for rates being quoted on market values      this rounds towards £1 billion.  Grossed up for total debt the notional (not actual) cost might be £1.5 billion
  • On index linked stock the rate my be 7.5%. That implies a cost for the month of around £3 billion.
  • The combined real cost is unlikely to exceed £4 billion, and maybe £4.5 billion including on government owned debt.

The claim is, however, that the debt cost in the month was £8.1 billion.

I accept that my calculations are assumption laden. I have also erred on the side of caution. should add that this  is completely normal in any form of accounting review of this sort where plausibility of a reported figure is being considered. Given that caveat, I still find the reported figure for interest incomprehensible.

The figure would imply an annual cost of nearly £100 billion. On market prices of gilts, gross (i.e. including government owned debt) that would imply a rate of return of 3.8% overall, and that is not happening, I suggest.

I have read, carefully, what the ONS has had to say on this issue. What seems most likely is that the figure reported is not a charge for December alone, but a sudden recognition that they had not included an appropriate sum to allow for the impact of increasing inflation on interest costs in earlier months. In effect, they were playing catch up, but not disclosing the fact. If so, the cost is much lower than implied during the month, and ONS accounting needs to be improved and greater candour in disclosure is required. As it is, the accounting is not good enough.

I will be asking the ONS to justify this claim, because I cannot.