An Analysis of the Pensions Regulator’s Review of impact on DB landscape following (the) LDI Episode: Part 4

Iain Clacher and Con Keating

This is the fourth, and last (for the time being), of our blogs concerned with TPR’s Report on LDI to Work and Pensions Select Committee. We will produce one final commentary when the ONS has published the Financial Survey of Pension Schemes (FSPS) for 2023. In time, when the ONS publishes their 2022 funding/liability survey results we may do a separate blog on this, but we have no firm idea when that might be as it is a new publication. The FSPS does not capture liabilities. The earlier blogs may be found from here:

A Continuing Commentary on the Regulator’s LDI Report (Keating/Clacher)

Several of our correspondents have drawn our attention to the language used by the Report. The use of the word “episode” to describe the events of 2022 and beyond was frequently cited as concerning as it feels like it diminishes the severity of what actually happened – an intervention in the gilt market by the Bank of England to maintain financial stability resulting from the use of LDI is not adequately captured by the word “episode”. We also have a particular concern in this regard with the use of “on-risk” to describe schemes which did not pursue LDI fully:

“24. Our modelling confirms that those schemes which were “on-risk”, i.e. those with higher levels of growth assets and lower levels of hedging, observed the greatest improvements over 2022. This is because these schemes have benefited from the significant fall in the value of liabilities linked to higher gilt yields whilst their asset values remained resilient.”

[Emphasis added]

Given the returns of gilts relative to other assets during 2022, risk here really is in the eye of the beholder. In this context, we are also concerned that for much of the decade prior to 2022, the returns to index-linked gilts, a key ingredient in LDI, was an assured loss when held for the long-term.

By this definition, Local Authority Schemes would have all been “on-risk”. Had the universe of DB schemes followed that example, private sector DB schemes would have lost just 8.4% of their assets, assuming the returns of the various asset classes were unchanged, which they most likely would not have been. That is a loss of £152 billion as measured by the ONS or £150 billion as measured by TPR, rather than the figures reported elsewhere in these blogs of £591 billion (ONS) or £425 billion (TPR). If scheme funding ratios had been 100% at the end of 2021, they would have improved to 136.8% at the end of 2022.

The Report contains the following description of TPR’s modelling methodology. It is unfortunate that the descriptions supplied are insufficient to replicate their models.

“65. The key assumptions underlying our model include:

o Trustees do not make any changes to the funding methodology/approach due to changing market conditions (or any other reason), when adjusting liabilities;

This is far from reality. The period under review was one of extremely substantial market change.

o Trustees do not take any management action in regards of buying/selling assets or changes to the investment strategy over time;

Trustees were selling assets to meet collateral calls throughout the year and much rebalancing took place. This, together with the earlier bullet, defeats the very purpose of the Report. These models cannot be expected to deliver the analysis requested by the Work and Pensions Select Committee.

o Discount rates are set with reference to a premium above gilt yields;

Using a fixed spread above gilts introduces a systematic bias as rates change. We illustrate this below, in Table 1, for a fixed margin of 2% above gilt yields of 1% and 5% for a discount function of twenty-year term. What is clear is that the margin dominates the discount function with gilts rate at 1%, making up 59.8% of the total discount, while with gilt rates at 5%, it is merely 16% of the total discount.

Table 1: Fixed margins and gilt levels

o Liabilities are adjusted in line with changes in gilt yields, market implied inflation and longevity expectations over time; and

Trustees are in fact required to set the assumptions used in their valuations at a prudent level relative to these. It would be interesting to see some empirical analysis of the relations between these ‘market’ rates and those prudent assumptions actually used by Trustees and their scheme actuaries.

o Assets are adjusted in line with market indices, plus deficit repair contributions.

It is worth noting that dealing costs are unaccounted for in market indices, and 2022 was a period with a significant amount of trading over and above what we would normally expect.

66. There are many more simplifications and approximations in the methods we use to estimate

aggregate and individual funding positions, compared with the more robust calculations carried out for formal valuation and recovery plan reporting by scheme actuaries and trustees. Additionally, the greater the magnitude of change in market conditions, the less reliable the simplified method and data will be in illustrating the impact. It should be noted that this is not a TPR-specific issue, but a global actuarial issue when using the approximate ‘roll-forward’ methodology to estimating assets and liabilities at alternative dates.”

[Emphasis added]

In light of the caution highlighted above, it is most surprising to see references to the use of PV01 in TPR’s models, for example:

“202. To determine the amount of interest rate hedging that schemes have undertaken these are calculated using the “PV01”. PV01 is the present value of the change in the value of the liabilities and the value of the assets from 1 basis point movement in interest rates.

203. Where the PV01 of the assets and liabilities is the same then this means that schemes will be 100% (of fully hedged) against interest rate movements. In other words, the asset value and liability value will move identically to any increase or decrease in interest rates (gilt yields) such that the funding level (A/L) is unchanged.”

PV01 is a measure of local change, that is one basis point in yield. When the change is of the order of the several hundred basis points as seen in 2022, it will result in gross errors.

Having the PV01 of assets and liabilities equal will only ensure equality over larger movements in their prices if the rate of change of the PV01 is also the same for both assets and liabilities.

There are similar concerns about the use of IE01 (Inflation Expectations).

The Report contains a brief discussion of the “Discounted selling of assets”.

“157. For some schemes to meet their collateral calls at short notice, they may have been required to sell assets at a discount relative to their recorded market value of that asset.

158. In general, we would expect such assets to have been less readily tradeable than more liquid assets. Furthermore, it is also noted that some discount may have applied to gilt sales, whereby the BoE purchased gilts at lower than market price (in order to ensure the liquidity of the gilt market).”

The Bank of England reported a profit of £3.8 billion on the sale of the assets purchased during their intervention. This was an explicit loss realised by schemes. The greater issue was that the bid-offer spread in conventional gilts widened almost fivefold, and quotes were good only for small amounts. However, the greater problem was experienced in index-linked gilts which showed ‘gapping’ in prices – for example, the 1/8% ILG due 2068 was offered at 50% and then traded at 44.5%. On several days during the crisis, price quotes were more indicative than tradable.

Investment grade corporate bonds, which were the subject of much selling by pension schemes, saw  yields, as measured by the S&P investment grade index, rise from a yield of 5.35% on September 21^st to 7.00% on October 11^th. This implies a price change of around 9%. It is worth noting that they have never revisited these higher levels since.

There were also reports of some selling of illiquid unlisted investments, such as private equity, at discounts to year-end valuations of 30% – 40%. However, the total volume of these transactions appears to have been relatively small, perhaps £3 – £5 billion.

After a rather strange illustration, this section ends with:

“162. Therefore, the extent of any loss observed where assets were sold at discount will depend on the size of the discount provided, as well as clearly both the level of hedging and the level of leverage adopted.

This is simply incorrect; the extent of any loss will depend solely on the amount sold and the discount applied.

163. We do not have any data regarding the extent and scope of discounted sales, and so the scheme funding figures from our internal modelling make no allowance for where this may have occurred. Furthermore, it is also difficult to make any estimates regarding such potential losses, given it will be very scheme-specific depending on levels of liquid assets, levels and types of illiquid assets, or whether schemes reduced the level of hedging instead.”

[Emphasis added]

This source of loss to schemes seems likely to have been relatively small, probably in the range £10 – £15 billion, but its absence from TPR’s modelling means that they will understate asset declines by this amount.

The report continues with consideration of “Loss of hedging”. Our impression is that this is a topic which is far more spoken about than actually happened, but we have only anecdotal evidence, as there is an absence of schemes reporting this, to support that view. From our analysis of individual company accounts, we have seen a good number of schemes selling growth assets to maintain their LDI positions. So, this is yet another area where we would have hoped that TPR’s report would have allowed us to gain a fuller insight into what has happened, but sadly it is absent from the report.

The Report correctly states:

“166. Where schemes did reduce the level of hedging, the subsequent impact on scheme funding will depend on various factors, including the date the hedge was reduced, the amount of any reduction, the date the hedge was re-introduced and the movement in financial markets over the period of time for which the hedge was lost. Indeed, the key point here is that it is what happens to gilt yields during the ‘under-hedged’ period that causes any potential funding gains or losses, rather than the act of replacement of the hedge itself.

167. In particular, if gilt yields rise following a reduction in hedging this would lead to an improvement in funding, whilst a fall in gilt yields will lead to a deterioration in funding (relative to the position if the hedge had been retained).”

Then the Report goes on to describe a complex, albeit stylistic model, which includes an impressively complicated chart, of the consequences of loss of hedging. Of course, without knowing the level of the incidence of such losses, no estimate can be made of the total cost of lost hedges. We have the impression that this model in intended compensate for the fact that TPR as yet cannot answer this question based on scheme returns and their current data and analysis, which as we have said elsewhere is concerning in the extreme given the time lapse between the crisis and this report.

The final section offers:

“173. In due course valuation submissions and annual scheme return asset data will reveal the extent of changes to funding levels and asset allocations following September 2022 and will be incorporated into our modelling following these submissions. However, this is expected to play out over several years depending on the dates of the next triennial valuation.”

This is actually an acknowledgement by TPR that their published statistics are at best incomplete.

The Report contains a section covering developments in 2023, “Changes to Scheme Funding”. We will not discuss this here, preferring to leave this to a commentary later in the year, when the ONS will have published its releases of the Financial Survey of Pension Schemes for June, September and December 2023, and (we hope) perhaps its work on funding and liabilities through 2022. We will though offer the comment that this section makes no mention of the continuing after-effects of the LDI crisis which seem by our reckoning to have persisted at least into the June 2023 quarter. We will also note one apparent anomaly. TPR reports scheme assets at December 2023 as £1,518 billion but the PPF reports them as £1,428 billion.

The body of the Report ends with “Closing remarks”.

“184. Whilst the situation in late September 2022 / early October 2022 bought into sharp focus the size and scale of DB pension scheme overall investments in leveraged products for gilts (LDI), the movement in gilt values (and yields) led to a significant improvement in scheme funding over 2022, due to the larger reduction in liabilities than that of asset values.”

[Emphasis added]

The central argument of this series of blogs has been that any improvement in scheme funding has actually been marginal rather than ‘significant’.

“185. TPR continues to enhance the data that we collect, particularly in respect of leveraged investments via the DB scheme return and for scheme liabilities through proposed changes in scheme funding regulations. This will enable us to improve our ongoing monitoring and evaluation of scheme funding in the future.”

This is welcome, but does it not raise questions as to the degree of knowledge and understanding possessed of a method of scheme management, LDI, which historically, TPR actively promoted.

“188. The precise impact of the events of 2022 for the whole universe will not be known for several

years, as each scheme goes through their tri-annual valuation process, and the ultimate impact on scheme funding becomes known. The first schemes required to complete valuations following September 2022, i.e. those with a valuation date of 31 December 2022, are due to complete these shortly. TPR will start to receive valuation submissions from those schemes required to produce a recovery plan. From a regulatory perspective, we will continue to engage with those schemes, on a scheme-by-scheme basis, where valuations pose the greatest risk to savers.”

In other words, it will be at least three more years before TPR have full coverage of the effects of the period of rising rates and their effects on schemes, and that of course will be confounded to some degree by subsequent developments.

Final thoughts

The Work and Pensions Select Committee recommended that:

“DWP should work with TPR and the PPF to produce, by the end of 2023, a detailed account of the impact on pension schemes of the LDI episode. This should:

1. look at the impact on funding levels, detailing how the value of their assets and liabilities changed, showing the results disaggregated by whether the fund used LDI and, if so, whether in a pooled, segregated or bespoke arrangement; and
2. include analysis of the factors which contributed to scheme funding improving or deteriorating, including the role played by LDI strategies.”

We leave it to readers to decide, in light of our various analyses whether TPR has addressed this recommendation adequately.