The commentary in the embed is a 17,000 word – 33 page – document with no pictures, jokes or cheap jibes. In short – it is written for people who have a serious interest in the governance of DB pensions in the UK. If you are interested in an alternative view from that peddled by the Pensions Regulator for the past 10 years then you should download and read it , because it challenges the orthodoxy that has seen Britain’s DB pension system turn from its “great economic miracle” to an “albatross around Britain’s corporate neck.
You can download this document as a PDF using this link
The document should be of interest to the following
- Regulators – including TPR, FCA and PRA
- Pension Trustees
- The Sponsoring Employers of DB pension Schemes
- Insurers looking to buy-out DB liabilities
- Those looking to consolidate DB pensions
- People in DB schemes with a mind to their likely future benefits
- The PPF
- The DWP and Treasury and MPs on the Work and Pensions Committee
- People with an interest in improving the productivity of the UK economy
- People with an interest in providing adequate retirement income for the UK’s elderly population.
I appreciate that will exclude many people who aren’t working in such organisations and have little interest in productivity or retirement outcomes.
Why does this critique matter?
There are two important issues that Keating and Clacher bring to our intention
- The discrepancies between reported data on pension assets between ONS, TPR and PPF – these are important as regulation and policy is data – driven.
- The intention of the impending DB funding code (to be delivered by April 2024) which unless subject to a rewrite from drafts, is likely to be a roadblock to the Mansion House reforms
Since the final draft of the DB Funding Code is yet to be published (expected January) , I am not going to comment further on that.
But I am publishing Appendix III of the document (which I have read in its entirety) as it explains why the PPF and TRP seem to have diverged in their assessment of the funding of DB pension schemes.
If we don’t have the true data , how can we learn its lessons? Even more importantly, how does dodgy data mis-shape future regulation?
The authors of the rest of this blog are Iain and Con.
Why the PPF and TPR can’t agree on the impact of LDI
joint authors- Dr Iain Clacher and Dr Con Keating
On December 13th the committee published a letter, dated 6th December, from Oliver Morley, the outgoing CEO of the PPF. The annual PPF Purple Book was published on December 6th and the letter uses a small extract from it.
The letter is a series of responses to questions posed to the PPF by the committee.
Among these questions posed are:
- You told us in April that a minority of schemes may have seen their funding position deteriorate following the LDI episode of September 2022 and there are several reasons as to why this may have happened which wouldn’t be captured in your estimate. We would be grateful for further detail on work since April to understand the position better, the number of schemes that have seen their funding position deteriorate, by how much, and the extent to which different factors have contributed to that;
- How much of the £400bn loss in asset value as a result of the LDI episode you would expect to see restored if interest rates rose again, and an explanation of how this would happen?
4. In response to a question on why your estimate of the reduction in the value of DB assets over 2022 differs from that of the Office for National Statistics (ONS), you said the data was collected for
The letter from Oliver Morley contains a truncated version of figure 2.4 from the 2023 Purple Book, which we reproduce in full below:
In response to the question:
“You told us in April that a minority of schemes may have seen their funding position deteriorate following the LDI episode of September 2022 and there are several reasons as to why this may have happened which wouldn’t be captured in your estimate. We would be grateful for further detail on work since April to understand the position better, the number of schemes that have seen their funding position deteriorate, by how much, and the extent to which different factors have contributed to that.”
The letter states:
“The data that we use in the annual Purple Book and monthly 7800 Index, which looks at the whole DB universe, is collected by TPR from scheme annual returns. Whilst TPR collect this data annually it is based on schemes’ most recent s179 valuations and in many cases, as schemes are only required to complete a s179 valuation every three years, these will not be current.
For example, the latest data provided by TPR from schemes’ annual returns includes valuations that have been updated since September 2022 for only 15 of our 5,051 scheme universe, as shown by the following table which is included in the Purple Book 2023:”
The letter then contains a truncated version of figure 2.4 above, containing only the numbers of schemes and the dates to which they apply. This text contains a further relevant footnote:
“We experience a further lag in receiving this data as schemes have up to 15 months to submit their valuation after the valuation date and the results are only reflected in the following 31 March scheme return”.
In other words, there are just 15 scheme reports which are close to being as current as those in the ONS survey sample of 614 schemes. We do not know the response rate to the ONS questionnaire but believe it to be typically about 85%.
We will make some observations on this:
The most current sample has the lowest funding ratio of all schemes in the universe (121.4%), far lower than the aggregate reported (134.3%). The highest (139.4%) is in the sample of 1,630 schemes dating from the period April 1st 2020 to March 31st 2021.
If we assume that these 15 schemes are in fact representative of the overall PPF universe, and use the PPF estimate of overall liabilities, with this funding ratio, the total assets of all schemes would be £1,269.5 billion. The latest ONS Financial Survey of Pension Schemes reported assets of £1,244 billion.
The letter continues with:
“It’s important to note that, even when we obtain data for the other 5,036 pre-1 October 2022 schemes, it will not be possible to form a meaningful estimate of how much of the funding changes arise from the LDI market disruption. This is because there are many factors that affect the rolled-forward assets and liabilities when moving to a new dataset. Principal among these are:
- changes in asset allocations/investment strategies between old and new datasets, to the degree that different asset classes have performed differently over the periods under consideration;
- any asset outperformance/underperformance between section 179 valuations relative to the market indices that we use for our roll-forwards;
- the performance between valuations of any unfunded hedging arrangements that are not captured in schemes’ disclosed asset allocations (for example interest-rate swaps);
- contributions made into schemes by both employers and employees between valuations;
- benefits paid out of schemes between valuations. Since pension retirement terms are not cost neutral when measured on a section 179 basis, commutation, early or late retirement will cause a surplus/deficit to arise.”
These are all valid points, but they arise from the staleness of the data inputs being used; the ONS sample, given the fact it is more contemporaneous, should not suffer from these problems. The answer does not, however, answer the question with respect to schemes which saw their funding position deteriorate. As we are not in possession of the full distribution of schemes, we also are not in a position to answer that question, though in our review of the other responses to questions, we are able to offer further insight.
As a prelude , we present a table below showing the evolution of some basic market metrics.
We show in the blue section, the yields and prices of twenty zero coupon gilt, as well as the aggregate decline in price from the prevailing prices in December 2021. It is also worth noting that the quarter to June 2022 saw rate rises and price declines almost as large as those to end September 2022, the beginning of the crisis. Note that from the end of September 2022, there was no clear gilt yield trend. In fact, looking at the discrepancy between ONS and PPF asset figures (the ochre section of the table above), we see the largest increase in the discrepancy of £136 billion in the April – June 2022 period. Half of the final discrepancy was evident by the end of June 2022. Note also that at the end of September only 61.7% of the final discrepancy was evident. Moreover, it is also clear that this growth in discrepancy is unrelated to discount rate movements.
From the simple repo leverage exposure (Green section) it can be seen that schemes in fact increased their gearing until the end of September 2022. This is even clearer if we consider the notional amount of exposure, that is the amount of repo divided by the price of the twenty-year gilt. This is shown below:
From this table, the small decline in the monetary value of repo between June and September was smaller than the declines due to price movements, and that the notional repo exposure was sold down after the September episode. At the end of this period, the notional exposure was slightly higher than at the beginning of 2022.
This is also evident in the net interest rate swap exposure;
By this measure, IR swap exposure resulted in total losses of £41 billion (to end September 2022) before there was any meaningful closure of these exposures. It appears that some 25% of IR swaps were unwound in the six-month period to March 2023.
We can offer a further and fuller analysis of the realisation of swap losses and the level of new cash committed if that would be of use to the committee’s investigation.
If we assume the underlying for the swaps was a 20-year fixed/floating swap that pays libor and receives gilt yield, we can also look at cash in and outflows due to swap realisation or termination as presented in the table below.
As the table shows, cash went into swaps in Q1 and Q3, totalling £51.1 billion; the £15.3 billion of Q1 2022 is cash margin from counterparties, while the Q3 £35.8 billion is new money subscribed to meet margin calls. Swaps were closed or realised with losses totalling £102.6 billion. Q2 also accounted for 44% of the remarkable growth in the discrepancy that quarter and resolving the mess after September 2022, saw £22.8 billion realised in Q4 and £20.3 billion in Q1 2023.
For completeness, the value of pooled LDI funds held by schemes fell from £231 billion to £168 billion after £51 billion was paid into them in the period. This is a loss of 49.4% (£114 billion) of the original investment but should not surprise, given the levels of leverage which were being used.
The questions and answers
Reverting to the PPF letter: (our comments are in red type)
“How much of the £400bn loss in asset value as a result of the LDI episode you would expect to see restored if interest rates rose again, and an explanation of how this would happen?”
“As set out above, the data we use when producing analysis of the whole DB universe is collected by TPR from scheme annual returns and is based on schemes’ most recent s179 valuations. However, schemes are only required to complete these every three years. In the absence of more up to date data on schemes’ asset allocation we aren’t able to provide an answer with any meaningful precision”.
This is true of the PPF data, but with the ONS data, we can make some first passes at estimating the losses unaccounted for in the PPF data and the extent to which they might be recoverable.
“However, regarding the mechanism by which a rise in rates would impact assets, it would actually be a reduction in interest rates that caused scheme assets to increase. For physical holdings such as fixed interest government bonds, the price is inversely related to the yield, so when the price of a bond rises the redemption yield will fall. For interest rate swaps, also commonly used to hedge interest rate risk, scheme trustees typically receive a fixed rate of return on a notional amount of “principal” and pay a floating rate that is linked to short-term interest rates. So if short-term interest rates fall, the pension scheme continues to receive a fixed rate from its counterparty to the swap, but pays out a lower rate, hence the present value of the arrangement will increase from the perspective of the pension scheme.
This really does not answer the question.
“In response to a question on why your estimate of the reduction in the value of DB assets over 2022 differs from that of the Office for National Statistics (ONS), you said the data was collected for different purposes—in your case, the potential impact on the PPF meant it was important to have information on funding levels across the PPF universe.
We understand that your estimates are based on a roll-forward methodology using data from scheme returns to TPR, whereas those of the ONS are based on a direct survey of a smaller number of schemes. Is the divergence between the two figures in 2022 exceptional, or have there been significant divergences in previous years? In addition, to what extent would it improve your assessment of the potential risks to the PPF if your estimates were based on more recent data?”
The response begins with:
“The divergence we’ve seen between the ONS survey and the 7800 index has been exceptional this year, however, it isn’t overly surprising as the ONS sample post-dates the LDI market disruption, whereas our data source pre-dates it. We are clear in our 7800 index publications that we don’t hold enough data to capture the structural changes to asset allocations, nor to capture changes to in any leveraged LDI portfolios (these factors have been particularly pronounced since March 2022) and, as a result, the impact on assets will often be less accurate than the ONS survey. As both measures use different methodologies. it is not possible to reconcile their outputs, and so, in the absence of more up-to-date scheme return data it is hard to know precisely what impact that would have on our view of wider scheme funding.”
It is clear that with more recent data, we can gain significant insights into schemes’ actual performance, for example, as we have shown above.
“With regard to understanding the risk to the PPF balance sheet, where we consider that a scheme sponsor has an elevated risk of insolvency we will proactively engage with scheme trustees, so that we understand those situations on a case-by-case basis.”
The PPF is faced with two issues not discussed. The general increase in sponsor stress arising from the rise in rates – see Appendix II above. There should be a second and immediate concern. If the ONS figures are correct, and there are many indications they are, then the level of scheme funding is far lower, by some £196 billion, and the PPF exposure to schemes in deficit would be more like £100 billion than the £2.2 billion, £5.7 billion or £8 billion that the PPF variously reports.
“You said that if schemes invested in assets whose value changed in the same way, then a scheme’s ability to meet the payments was unaffected. How are you able to assess the extent to which schemes have effectively matched their assets and liabilities?”
“We use data on asset allocation provided to us by TPR from schemes’ annual returns. However, as we note above, in many cases this data is not current. In addition, this data does not completely capture the impact of leveraged exposure, however, we understand that TPR plan to collect richer data on leveraged exposure in the future.”
It is actually in the overwhelming majority of schemes’ where the PPF data is not current. From our other empirical work, it would appear that schemes were on average between 75% and 80% hedged – though there were some schemes which were very highly geared, with some schemes we have seen being 124% hedged. One point to note is that schemes which were not utilising LDI would typically own some bonds as part of their traditional diversified equity/bond asset allocation with 20% – 40% allocated to bonds not unusual among these schemes.
We can make some estimates of the degree to which the losses experienced might be recovered should rates return to their previous levels. There is the trivial £3.8 of profits made by the Bank of England which will not be recovered. Only about half of the £114 billion lost by pooled funds can be expected to be recovered given their deleveraging and lower gearing post-crisis. Of the £51 billion lost on IR swaps, £30 billion might be expected to be recovered. Of the losses on repo, approximately 75% should also be recoverable. Put another way, if we consider all of the £196 billion discrepancy as LDI specific losses, we should only expect to recover only some 50% – 65% of these losses. This is conditioned on rates falling to the previous lows, and index linked gilts recovering to their old highs offering yields as low as RPI – 3.5%.
It is clear that the costs of LDI since rates began rising have exceeded the gains made by schemes previously. This has also been true of the Bank of England’s QE asset portfolio. It is also most unlikely that we will see such low rates again in the remaining lifetimes of schemes, and there is no indication from the Bank of England that we will be going to back to base rates of interest as low as 0.1%.