Philip Bennett’s DB funding code response

Note: I am responding to the Consultation Document in my personal capacity.

Executive Summary

This response to the Consultation Document is limited to examining critically three key aspects which underline the Pensions Regulator’s preliminary conclusions for a long term funding objective in a quantifiable (or prescriptive) form to be achieved by defined benefit schemes by the time they are significantly mature (as defined in the Consultation Document).
The three key aspects examined in this document are:

2.1 reliance on the GAD modelling as the key foundation for the proposed quantified LTO.

2.2 whether a quantified LTO will encourage, or further encourage, procyclicality in the investment of DB Scheme assets and the creation/enlargement of systemic risk in DB Schemes.

2.3 cost and value for money relative to additional member protection conferred in requiring DB Schemes to achieve a quantified LTO by the time they are substantially mature (as defined in the Consultation Document).

Here is an impact scorecard, based on the analysis set out or referred to in this response to Consultation Document, on the preliminary conclusions reached in the Consultation Document at paragraphs 638 and 639 and the articulated desire to have a more or less prescribed set of valuation assumptions and investment strategies for the long-term funding objective (the “LTO”) and for determining technical provisions.

Impact scorecard

	Issue	Positive or negative impact	Comment
1.	Improvement in member benefit security	Positive	1.1 But, assuming the journey plan is achieved, which (depending on the time period to maturity 14) takes 15 to 25 years? 1.2 Very poor value for money -see 6 below?
2.	Measurability to facilitate comparison of DB Scheme profiles from a regulatory perspective by reference to the measurement tools used	2.1 Makes comparisons (assuming the measurement tools are useful and valid) easier. 2.2 Facilitates “box ticking”. 2.3 Problem if measurement tools are not fit for purpose and value destroying.	2.1 If adopted, enables a “fast track” valuation review approach. 2.2 Severe negative impact -see 4, 5 and 6 below relative to intended improvement in member benefit security?
3.	LTO based on: 3.1 robust modelling? 3.2 Pensions Regulator decision taking based on understanding of limitations of modelling? 3.3 decision taking of Pensions Regulator in adopting LTO informed by cost (and value for money) of reducing risk of member benefit reduction relative to level of improved protection of member benefits?	3.1 doubtful 3.2 unknown 3.3 unknown. But, if not taken into account, decision taking process lacks credibility and could be suggestive of governance failure or regulatory capture.	3.1 In terms of transparency of decision making and good governance, it would be helpful for the Pensions Regulator to publish more detail about its decision taking process in adopting its preliminary conclusions in paragraphs 638 and 639 of the Consultation Document. 3.2 In particular, it would be helpful to understand what measures have been taken to avoid the danger of “group think”.
4.	Encourages procyclicality?	Yes.	4.1 Encourages the purchase of gilts and bonds in circumstances where the gilt and bond markets are affected by central bank quantitative easing programmes (ie the central banks are rigging those markets). 4.2 Compatible with DB Scheme trustees’ prudent person and other investment duties if you invest with the ex ante knowledge of a negative real investment return?
5.	Does the proposed LTO (and associated journey plan) create/enlarge systemic risk?	Yes	5.1 Given that the future is unknowable (and particularly over very long time horizons), it is poor regulation to encourage all DB Schemes to adopt similar investment strategies (herding). 5.2 A case of putting all of your eggs in one basket. 5.3 Consider, in particular, impact on DB Schemes which use a 3 x leveraged LDI strategy if long term rates rise sharply.
6.	Does the proposed LTO (and associated journey plan) provide value for money relative to the reduction in risk to member benefits achieved in terms of: 6.1 overall cost (mostly borne by employers), and 6.2 cost borne by the taxpayer?	6.1 No cost assessment has been published by the Pensions Regulator but ball park figures for additional cost could be in the region of £100 billion to £120 billion as at 31^st March, 2019. 6.2 A ball park figure of cost to the taxpayer could be in the region of £20 billion to £24 billion as at 31^st March, 2019. 6.3 The cost of achieving the LTO for all schemes immediately as at 31^st March, 2019 would have been, on the Pension Regulator’s figures in the Consultation Document, £555 billion.[1]	6.1 These figures are deserving of full scrutiny[2](and it seems odd that the Pensions Regulator did not include its assessment (including assumptions and methodology) in the Consultation Document). 6.2 They may, of course, have increased since 31st March, 2019. 6.3 But, they also raise the question whether there are alternative ways of reducing the risk of reduction of member benefits. 6.4 The risk of reduction to member benefits in the 15-25? year journey plan to the LTO is not modelled or estimated nor is the counter factual of doing nothing estimated or modelled. 6.5 If you join up the dots, it is surprising that HM Treasury does not require an approach by the Pensions Regulator to its current and proposed valuation methodology to demonstrate value for money for the tax payer relative to the improvement in the level of protection for member benefits.

I have suggested an alternative approach to seeking to improve the level of security for member benefits on what should be a substantially more cost effective basis both to employers and also to the taxpayer.
It must not be forgotten that every pound of employer contribution will, for a taxpaying employer, attract a corresponding tax deduction from profits with the taxpayer footing the bill for c. 20p in the pound for every pound of employer deficit make up contribution[3].
More generally, in a time of rigged gilt markets (via quantitive easing), a regulatory approach which promotes buying gilts or other bonds to achieve the LTO (and the associated post LTO investment strategy (at negative real rates of return) heavily based on modelling:

“to protect member benefits, it is important to maintain an investment strategy that is highly resilient to risk”[4]

appears to have some analogy with this year’s A level results and the pursuit of the target of no grade inflation via the use of algorithms (or models) in not looking at the bigger picture.

[1] See Part V, Section A4 of this document for more details.

[2] See Part V, Section C3 of this document for more details of the calculation sources.

[3] See Part V, Section C4 of this document for more details.

[4] The Consultation Document, paragraph 639.

Part I: Introduction

Scope of this response
This response to the Consultation Document is limited to examining critically 3 key aspects which underpin the Pensions Regulator’s preliminary conclusions for a long-term funding objective (the “LTO”) in a quantifiable form to be achieved by the defined benefit schemes (“DB Schemes”) by the time they are significantly mature (as defined in the Consultation Document).
The 3 key aspects, which are examined in Part III, Part IV and Part V, are:

2.1 reliance on the GAD modelling as the key foundation for the proposed quantified LTO.

2.2 whether a quantified LTO will encourage, or further encourage, procyclicality in the investment of DB Scheme assets and the creation/enlargement of systemic risk in DB Schemes.

But, before examining those 3 aspects, Part II provides some relevant background information and context to support the analysis and conclusions reached in Part III, Part IV and Part V.
Part VI suggests a possible alternative approach to achieve the identified objective of providing additional protection for member benefits in DB Schemes at a cost which could be both affordable and provide better value for money for employers, tax payers and other stakeholders.
Providing a framework for enhancing the already high level of security for member benefits in an affordable and value for money manner is a very difficult problem to solve
As is acknowledged by the Department for Work & Pensions White Paper “Protecting Defined Benefit Pensions”[1] (“White Paper”):

“Regulating the Defined Benefit sector is a challenging job”.

The Consultation Document (and all of the work that went in to producing that document) illustrates the complexity of the issues in relation to funding of DB Schemes to enhance further the security for member benefits against costs to the employers, the tax payers and other stakeholders.
As is acknowledged in the White Paper:

“In the natural course of business some employers will become insolvent. While we are putting in place measures to increase the protection of members’ benefits, the proposals in this White Paper cannot prevent insolvency. Nor is it always in the interest of members of a pension scheme for the Regulator to force a company which is already struggling, to pay substantial pension contributions that risk worsening the situation. In the long-term, the best protection for a Defined Benefit pension scheme is a strong and solvent employer which works with trustees to put the needs of the pension scheme on an equal footing to other business considerations.

In the event of employer insolvency, there is a proven system in place to reduce potential losses to members. The Pension Protection Fund was established in 2005 to pay compensation to members of eligible Defined Benefit pension schemes where the sponsoring employer becomes insolvent. In such circumstances, when schemes do not have sufficient assets to secure pension benefits at the compensation level or above, the Pension Protection Fund, which is funded by a levy on all Defined Benefit pension schemes, steps in.

The level of protection afforded to members is high, regardless of the level of underfunding in the scheme. Compensation is broadly paid at 100% of accrued benefits to those over the scheme’s normal pension age and 90%, subject to a cap, to everyone else. As at March 2017, around 130,000 members were in receipt of compensation. Another 110,000 deferred members were in the Pension Protection Fund but with compensation not yet in payment. This is equivalent to only just over 2% of all Defined Benefit scheme members. The Government notes with concern that even those attempting to highlight the impact of business failure on pension scheme members do not always recognise the high level of protection the Pension Protection Fund provides.

Whilst the safety net provided by the Pension Protection Fund is a key feature of the system, this Government will not stand for employers evading their responsibilities and relying on this protection, which could lead to other sponsoring employers, playing by the rules, to foot the bill.”[2]

(emphasis added)

I do not underestimate the difficulties and complexities involved. But I remain sceptical as to whether the attempts as proposed in the Consultation Document to quantify the unknown future and then to use that quantification as a regulatory tool do anything other than increase risk (including systemic risk) and impose extremely substantial costs on sponsoring employers (and also the tax payer via the tax relief given to employer contributions).
The comments in this document are intended to be constructive comments and, I hope, will be read in that light.
Some context for ease of reference:
The Consultation Document at paragraph 626 says:

“626. GAD’s base case analysis assumed a low dependency funding target based on a discount rate of Gilts plus 0.5% (approximating to 93% of buy-out cost before expenses).”

The Consultation Document, at paragraphs 638 and 639, sets out the preliminary conclusions of the Pensions Regulator as to the LTO to be achieved by all DB Schemes and the time by which that LTO should be achieved by each scheme:

“638. While the GAD modelling has its limitations, it nevertheless shows that an LTO can be set so that there is a high likelihood of a typical mature closed scheme funded at a low dependency level and invested on a low risk basis being able to survive on its own with minimal reliance on employer support.

In particular, the analysis shows that setting a discount rate of between 0.25% pa and 0.5% pa in excess of gilt yields for a scheme with a duration of 14 years would appear to fit the definition of low dependency. This is because there is a low chance of requiring any further support from the employer and a very low chance of that support being significant relative to the original size of the scheme. It also shows that, to protect member benefits, it is important to maintain an investment strategy that is highly resilient to risk.”

(emphasis added)

But what is not said in the Consultation Document is how much such an investment strategy which is highly resilient to risk costs relative to the counter factual of doing nothing or of alternative solutions for seeking to increase the already high level of protection for member benefits.
Given the Pensions Regulator’s correct focus on value for money in relation to DC Schemes, there is no discussion of value for money in the Consultation Document.
Questions

If anything in this document is not clear or it would be helpful to discuss any aspect of this document, please do not hesitate to contact me[3].

Part II: Some background and relevant factors

Introduction
The proposed approach set out in the Consultation Document to defining, in quantitative terms, the long-term funding objective is underpinned by the substantial modelling which has been undertaken by the Government Actuary’s Department as set out in their document of 14th February, 2020 (the “GAD Modelling Document”)[4].
However, we are now living in some very unusual times in a world of quantitative easing with:

2.1 UK gilts having negative real (ie. after inflation) yields at all maturities[5] (and also negative for some UK gilts on a nominal yield basis)[6], and

2.2 Eurozone Government bond yields having negative yields on a nominal basis[7].

We are in a situation where no-one knows how quantitative easing will unwind. It is, therefore, difficult to see how a model such as that used for the modelling undertaken by the Government Actuary’s Department can model the unknown with any credible degree of reliability.
Given that the purpose of quantitative easing, as outlined in more detail in Section B below, is:

to reduce interest rates (including on gilts and bonds),
it seems perverse to encourage pension schemes to aim to achieve an LTO which is substantially based on investing in gilts and other bonds.

Impact of quantitative easing in relation to pension fund valuation methodology
What is quantitative easing?

1.1 Quantitative easing occurs when a central bank creates money to buy bonds with a view to reducing interest rates.

1.2 Its aims are:

(a) to reduce the cost of borrowing for companies and individuals so as:

(i) to reduce the risk of insolvency/bankruptcy insofar as the debt obligations of those borrowers bear interest at a floating rate rather than a fixed rate, and

(ii) to make it cheaper for companies or individuals to borrow money which the central bank intends should lead to an increase in current consumption and stimulate current economic growth, and

(b) more generally to encourage consumption and discourage saving (ie. deferred consumption) so as to stimulate current economic growth.

1.3 To implement its quantitative easing policy, the Bank of England is a major buyer of fixed interest gilts.

1.4 In a usually efficient market, if you have more demand than supply, the price of the fixed interest bond goes up and the yield to the buyer of the fixed interest bond goes down.

1.5 But, perhaps unlike an efficient market, you have a major player in the market, the central bank, rigging the market.

How much quantitative easing has the Bank of England done?

2.1 In the period December 2009 to June 2020 the Bank of England has purchased £745 billion of gilts (ie, the aggregate of the amounts in 2.2 below).

2.2 This chart shows the buildup of the £745 billion:

Source: Bank of England: Available at this link accessed on 24th August, 2020: https://www.bankofengland.co.uk/monetary-policy/quantitative-easing.

Ed; Philip Bennett apologizes for an earlier version of this chart which erroneously referred to £2,400 bn. of QE.

Bank of England’s expectation of pension fund behaviour

3.1 Here is a view expressed by the Bank of England as to its expectation of pension fund behaviour in response to quantitative easing:

“Large-scale purchases of government bonds lower the interest rates or ‘yields’ on those bonds (the investopedia website explains more about bond yields). This pushes down on the interest rates offered on loans (eg mortgages or business loans) because rates on government bonds tend to affect other interest rates in the economy.

So QE works by making it cheaper for households and businesses to borrow money – encouraging spending.

In addition, QE can stimulate the economy by boosting a wide range of financial asset prices.

Suppose we buy £1 million of government bonds from a pension fund. In place of the bonds, the pension fund now has £1 million in money. Rather than hold on to this money, it might invest it in financial assets, such as shares, that give it a higher return. And when demand for financial assets is high, with more people wanting to buy them, the value of these assets increases. This makes businesses and households holding shares wealthier – making them more likely to spend more, boosting economic activity.”[8]

(emphasis added)

3.2 However:

if you use discount rates derived from gilt yields to measure the present amount of future pension obligations,
the lower the discount rate, the higher the present amount of those future pension payment obligations,
even though the pension payment obligations remain unaffected (leaving to one side the impact of a lower or higher rate of inflation on the revaluation of deferred pensions and on the level of increases to pensions in payment),
companies have to contribute more to their defined benefit schemes (so reducing the amount available to invest their business[9] and weakening the employer covenant), and
households saving for their retirement in DC schemes have to save more for their retirement (making those households poorer).

What impact has quantitative easing had on long-term interest rates?

4.1 The short answer would appear to be that quantitative easing has reduced significantly long term interest rates.

4.2 As to exactly how much, is difficult to establish precisely. But it is worth noting the conclusion from Bank of England Working Paper No. 466 published in October 2012 in respect of the first phase of quantitive easing (£200 billion – see 2.2 above):

“Overall our results imply that the Bank’s QE asset purchase had a significant and persistent impact on gilt yields. Our paper also provides evidence of local supply and duration risk effects consistent with the imperfect substitution, which has implications beyond the financial crisis for how we think about price determination in the gilt market.”[10]

(emphasis added)

4.3 The same paper, at page 40, comments:

“Analysis of the high-frequency market reactions to individual announcements on QE suggested that the initial impact from the announcements took time to be fully priced in and that the cumulative initial impact on yields varied significantly across the term structure, with the largest impact up to 120 basis points at the 15-20 year maturity.”

(emphasis added)

4.4 In July 2009, the long (30 year) gilt yield was 4.49%[11]. The first phase of gilt purchases in November 2009 was £200 billion. Since then, the Bank of England has purchased a further £545 billion of fixed interest gilts[12]. In July 2010, the long (30 year) gilt yield was 4.22%[13] . In July 2020, the long (30 year gilt) yield was 0.64%[14].

4.5 The number of basis points yield reduction per £200 billion of gilts purchased does not necessarily scale. But it is open for debate as to whether the Bank of England’s quantitive easing has reduced the long (over 30 year) gilt yield by as much as 350 basis points over the 2010 to 2020 period.

Has quantitative easing adversely impacted on DB Schemes?

5.1 It may be helpful to note the following evidence submitted in 2013 to the House of Commons Treasury Committee’s Inquiry into Quantitative Easing[15]:

“4. Defined Benefit pension schemes

QE has damaged UK defined benefit pension funds and their sponsors:

As the UK pension system is underpinned by gilt yields, artificially depressing longterm interest rates also depresses pension values, which imposes extra costs on companies sponsoring pension schemes, thereby reducing their earnings. The Bank’s distributional analysis report misrepresents QE’s impact on pension funds. It states: ”The Bank’s assessment is that asset purchases have pushed up the price of equities by at least as much as they have pushed up the price of gilts”. QE has, of course, pushed up gilt prices, but equity markets and corporate performance also depend on other factors. In fact, pension assets have not kept pace with gilts and the Bank’s assertions that equity price rises have offset liability increases also ignore market volatility.

An example of the volatility of markets for defined benefit pension schemes is shown in the Pension Protection Fund regular reports on the health of UK pension schemes.

In July 2011, the PPF deficits for UK pension schemes totalled £8.3bn.
Just six weeks later, the PPF deficits had increased to £117.5bn.
Then in July 2012 deficits had ballooned to £300billion.

This volatility of asset prices and pension deficits make falling gilt yields resulting from QE particularly problematic for companies already struggling to support their pension promises.

Indeed, traditional estimates suggest that a 1 percentage point fall in gilt yields leads to approximately a 20% rise in pension liabilities, but only a 6-10% rise in typical pension fund asset values. The Bank does admit that if pension funds were in deficit (as they almost all were in 2009) then QE will have worsened those deficits. However, it does not even attempt to quantify the actual effects on schemes. It also ignores the impacts on corporate UK. In fact, the damaging impact of QE on scheme deficits is another factor that may be undermining the effectiveness of QE itself.

Firms have had to put more money into their pension schemes, as their pension deficits have increased. This leaves less money to spend on expanding their businesses or creating jobs and mergers and acquisitions or corporate restructuring are hampered. Companies with pension schemes are also finding it increasingly difficult to access new bank loans. Some firms have even failed as a result of their pension problems.

QE is locking many defined benefit pension schemes in a vicious spiral:

Pension funds are becoming locked in a vicious circle in that they need money to support their business so they can earn more money to fund their pension scheme, but they cannot obtain more money because of the scheme itself. In addition, pension trustees with rising deficits often want to ‘de-risk’ by buying fixed income assets that more closely match their liabilities, or try to buy out their liabilities to reduce risk. But the more the Bank of England buys gilts, the more expensive – and unaffordable – it becomes for trustees to buy assets to reduce risk. If they compete with the Bank of England to buy more gilts, they drive gilt prices up more, which increases their deficits even further. For many schemes, this can be a ‘death spiral’. The Bank assumes[16] pension funds are or should be invested in assets that ‘match’ their liabilities so that falling gilt yields will not impact them. This misunderstands pension reality. Firstly, there are no matching assets. Index-linked gilts only partially match inflation risks. Secondly, there are not enough index-linked gilts available so pension funds could not match their liabilities as they would need more than the total gilts in issue. Thirdly, UK pension funds hold a diversified range of assets, so their assets have not kept pace with gilts.

The Bank’s complacency on this issue is troubling. Pension funds are meant to be long-term, but QE has distorted them over the short-term. A recent note from Morgan Stanley estimated that firms with the biggest pension deficits compared to their market capitalisation have underperformed the stock market by 28% in the past three years.”[17]

(emphasis added)

5.2 The point made in Baroness Altmann’s evidence about the imperfect relationship between the price of gilts and the price of equities is further supported, for example, by analysis done by Robert Shiller (albeit by reference to US markets):

“Although interest rates must have some effect on the market, stock prices do not show any simple or consistent relation with interest rates.”[18]

(emphasis added)

How will quantitative easing end?

6.1 The short answer is that nobody knows.

6.2 Some would argue that, as (or when) the economy recovers and inflation starts to rise, the central bank will:

gradually sell the bonds it has purchased back on to the market to damp down interest rate rises, or
hold the bonds to maturity.

6.3 Under that scenario, interest rates will (gradually?) increase and the prices of bonds which have already been issued will fall.

6.4 Others would argue that quantitative easing will ultimately lead to an increase in inflation (possibly a very substantial increase in inflation); perhaps as a result of a currency crisis with the rising cost of imports feeding through into inflation. In turn, at least based on conventional wisdom, this is likely to lead to an increase in interest rates to seek to improve the exchange rate, to curb consumption (and to seek to control inflation rising above the targeted level). Again, the price of already issued bonds would then fall.

6.5 Common sense suggests that the prudent pension fund trustee, who has not yet reached full funding on the proposed LTO (with technical provisions calculated on a gilts + 0.5% basis) should not be putting all of its eggs in seeking to achieve an LTO investment strategy (largely based on holding bonds of appropriate maturities):

(a) based on a market which is being rigged by a central bank, and

(b) where there is substantial uncertainty as to how quantitative easing will end.

6.6 In the next Section, I look at some points on the key assumptions in the GAD Modelling Document.

Gilt yields as the yardstick to measure the risk free rate of return:
Credit risk

1.1 So long as the UK Government retains control of its own currency, it has the ability to print as much money as is necessary to meet its debt obligations.

1.2 So, in nominal terms[19], it can be argued that there is close to no credit risk on a sterling denominated gilt.

1.3 Logic suggests that the prudent (or rational) long term investor:

will expect to be paid for the time value of his money in buying a gilt, but
if there is close to no credit risk, that such prudent investor, if investing in other asset classes, would expect to earn a premium over the “risk free rate of return on gilts to reflect, among other things, the associated lower credit risk.

1.4 However, a piece of history from 1918 shows that the credit risk is not necessarily zero. The manifesto of the Soviet of Workers’ Delegates from 1905 said:

“The government is on the brink of bankruptcy. It has reduced the country to ruins and scattered it with corpses. The peasants, worn out by suffering and hunger, are incapable of paying taxes. The government gave credits to the landowners out of the people’s money. Now it is at a loss as to what to do with the landowners’ mortgaged estates. Factories and plants are at a standstill. There is unemployment and a general stagnation of trade.

…

We have therefore decided not to allow the repayment of loans which the government contracted while it was clearly and openly waging war against the entire people.”[20]

1.5 The decree for the repudiation of debt was adopted by the Soviet government in February 1918.

1.6 This in an extreme outturn to hypothesise for UK gilts. But it is not inconceivable in the logical certainty sense.

The prudent long investor: protecting nominal amounts

2.1 The prudent long term investor, it can be argued, expects to obtain (at the time the investment is made):

a 100% return for the capital invested, and
a risk adjusted, as perceived by the prudent long term investor, return on that capital.

2.2 The expectation in 2.1 can, more accurately, be attached to the overall expected return on a portfolio of diversified assets.

2.3 In the case of sterling denominated gilts, which, as noted in 1 above, do not appear to have any material attaching credit risk, if the market is not being rigged, prima facie, the yield on those gilts may be expected to reflect the prudent long term investor’s perceived absence of credit risk.

The prudent long term investor: protection of investment against the effect of inflation

3.1 Logically, the prudent long term investor buying a gilt would expect to earn a rate of return on the gilt which provides:

a positive rate of return in nominal terms, and
a positive rate of return after allowing for expected inflation over the time period of the investment.

3.2 That said, if there is a greater demand for gilts through central bank quantitative easing, such that the return expected by the prudent long term investor referred to in 3.1 above is not met, the starting presumption is that the prudent long term investor would not invest in that particular asset class.

Has the prudent long term investor become imprudent?

4.1 Chart 1 below shows that all UK gilts as at 21st August, 2020 have a negative real return (at all maturities) in excess of 1.5% per annum.

Source: Yield curves as published by the Bank of England as accessed on 24th August, 2020 and available at this link: . https://www.bankofengland.co.uk/statistics/yield-curves

4.2 This raises a question whether the prudent long term investor:

(a) should continue to hold gilts and other bonds which have an average negative return in excess of -1.5%, or

(b) should sell those gilts and other bonds and invest the proceeds of sale in other asset classes (see further the Bank of England expectations referred to at B3 above).

Euro area yield curves as published by the European Central Bank

5.1 Although, currently, not directly relevant, Chart 2 below shows that Euro area yield curves for AAA rated Euro area central government bonds are showing a negative nominal return at all maturities.

Source: As published by the European Central Bank on 21st August, 2020 and available at this link:
https://www.ecb.europa.eu/stats/financial_markets_and_interest_rates/euro_area_yield_curves/html/index.en.html

5.2 In other words, the investor:

invests €100 in a AAA Euro area Government Bond,
enjoys a negative interest rate for the period that the bond is being held (ie. the capital amount invested is reducing), and
at maturity, receives back a guaranteed lower (nominal) amount than the investor invested.

5.3 An interesting feature is that the investor is prepared to invest for up to a 23 year term with a guaranteed loss of at least 0.1% a year in nominal terms. In other words, I lend you €100 for 23 years and you pay me back €97.7!

5.4 If the Bank of England’s quantitative easing programme starts to move the nominal yield on gilts into negative interest rate territory (as for the Eurozone), is pursuit of the proposed LTO by DB Schemes (along with the other asset allocation changes already in train as currently being encouraged by the Pensions Regulator’s derisking approach), a course of action which a prudent long term investor should be taking?

5.5 Of course, from a short term trading perspective, you could argue that:

if you believe that the effect of more quantitative easing will be to push more UK gilt returns into a negative yield territory, then
you would buy more gilts now and should yields fall further realise a significant capital gain.

5.6 An example of this is the 100 year bond issued by Austria in 2017 which, 2 years later, had delivered a 75% gain[21].

A floor to negative interest rates?

6.1 It may be that the next step in the Bank of England’s quantitative easing programme will be to reduce the nominal yield on gilts in to negative interest rate territory.

6.2 It may be argued that there is a floor to negative interest rates[22] in that the prudent investor can rent a bank vault and hold cash in the bank vault (so the floor equals the cost of holding cash in the bank vault and accepting the risk of the cash being stolen or destroyed).

6.3 The counter-argument is that the supply of bank notes can be constrained so that a buyer of banknotes will have to pay more than face value for the bank notes to counteract the effect of the negative interest rates. In other words, to misquote what a Prime Minister said nearly 53 years ago[23] “the £1 in your pocket would be worth more than £1”!

6.4 Whether such a move in depreciating the nominal amounts of savings held in bank and building society accounts by savers will prove politically popular is another matter.

6.5 Unlike erosion of purchasing power through inflation, it is very easy to see the impact each month of your savings being reduced by a negative interest rate.

Are there valid reasons why a prudent investor would not seek to invest to achieve a real return after inflation

7.1 In the strange world of quantitative easing, it is, perhaps, worth taking a sense check as to whether risk management techniques have run amok and are destroying value.

7.2 If:

(a) a pension fund investor is seeking to “match” the way in which:

(i) as at a valuation date the present amount of future pension payment obligations are measured in a valuation (ie the technical provisions)

(ii) by investing in assets where the market value of those assets moves in line with the discount rate used to determine the technical provisions,

then

(b) it may be argued that this is a prudent and rational investment approach.

7.3 Or it may be that the DB Scheme investor is being nudged, encouraged or forced via the risk averse stance adopted by the Pensions Regulator (and to be hardened via quantification and the LTO, in the proposed DB Funding Code) so as, in effect, in effect to have to invest in a manner that, ex ante, can be demonstrated to be value destroying (with the amount of value destroyed, being capable of ex ante quantification).

7.4 It is, perhaps, also worth noting that the more employers pay into their pension schemes to reduce deficits created by the lower and lower discount rates used, the weaker the employer covenant is likely to become.

7.5 That said, I fully accept that there needs to be an equitable treatment of the DB Scheme relative to the other uses of the available cash of the DB Scheme employer.

7.6 But the Pensions Regulator already has extremely wide powers[24] (to be further expanded by the Pension Schemes Bill 2020) to control cash leakage via abnormal dividends, via share buybacks or other value leaking transactions.

The starting position in investing DB Scheme assets is to invest them as a prudent person would

8.1 The starting position is that:

(a) scheme assets should be invested with the aim of achieving a long term rate of return which would enable pensions to be paid as an when they fall due,

(b) with an investment strategy which is appropriately diversified by asset class (a case of not putting all of your eggs in one basket as a way of managing risk and uncertainty),

(c) with the expected investment return on the assets necessary to pay the pensions as and when they fall due including a suitable margin for prudence, and

(d) to adopt an appropriate cash flow management strategy to reduce forced disinvestment risks at a time of market stress in order to pay benefits through holding a sufficient reserve, where necessary,[25] of cash or near cash assets corresponding to expected short term outflows to ride out market falls.

8.2 The DB Scheme in question is in the position to estimate the likely starting dates, duration and amount of pensions it has to pay by making assumptions as to:

(a) longevity, and

(b) inflation (in respect of revaluation in deferment and increases in payment albeit generally capped[26] at 5% or 2.5%).

8.3 However, do not lose sight of the fact that longevity assumptions can be:

(a) affected by factors that can be viewed as “favourable” to the scheme such as plagues, natural disasters, climate change and wars which reduce the longevity of scheme members, but

(b) affected by factors “unfavourable” to the scheme such as improvements in medical treatments and cures.

8.4 However, the key assumptions as to the expected future returns on particular asset classes, with the exception of a current holding of fixed interest gilts, fall into the educated guess category.

8.5 You could, therefore, reason that one investment strategy would be to invest 100% in a mixture of fixed interest and index linked gilts so that you would have largely eliminated interest rate and inflation risk (but not longevity risk).

8.6 A problem with that approach is cost. At the current time you end up in a position where the sum of the expected payments out (without discounting and excluding inflation related increases) is lower than the market value of assets the scheme holds in a gilt or gilts + 1.5% portfolio (see C4.1 above).

8.7 But, it is accepted, in the White Paper, that there is a balance to be struck between affordability and risk:

“In the long-term, the best protection for a Defined Benefit pension scheme is a strong and solvent employer which works with trustees to put the needs of the pension scheme on an equal footing to other business considerations.” [27]

No crystal ball to predict a future investment return

9.1 An actuary or investment consultant has no special crystal ball that enables that person to forecast with any precision what the long term returns will be on particular asset classes (leaving to one side current holdings of gilts which have a known redemption date and a known coupon).

9.2 For current holdings of other bonds there is a known maturity date and a known coupon, but there is credit risk which is unknown.

9.3 An economic scenario generator is not a crystal ball (and see further discussion in Part III, Section D).

9.4 Without intending to cause offence to actuaries, it is, perhaps, worth considering the following:

“Consulting actuaries are very good at making calculations. They are frequently terrible at making the assumptions upon which the calculations are based.

In fact, they well may be peculiarly ill-equipped to make the most important assumptions if the world is one of economic discontinuities.

They are trained to be conventional.

Their self-interest in obtaining and retaining business would be ill-served if they were to become more than mildly unconventional.

And being conventional on the crucial assumptions basically means accepting historical experience adjusted by a moderate nudge from current events. This works fine in forecasting such factors as mortality and morbidity, works reasonably well on items such as employee turnover, and can be a disaster in estimating the two most important elements of the pension cost equation, which are fund earnings and salary escalation.”

Source: “Memorandum regarding the pitfalls of pension promises” written by Warren Buffett in 1975 for the Directors of Washington Post Company and republished in the Berkshire Hathaway Annual Report for 2013 http://www.berkshirehathaway.com/2013ar/2013ar.pdf .

A clash between investment duties of DB Pension Scheme trustees and the proposed derisking approach in seeking to attain and implement the LTO in a quantitative easing world
Duty to act as a prudent person would when investing

1.1 The DB scheme trustees have a prudent person duty to discharge when investing scheme assets.

1.2 That duty comes from the IORP II Directive[28], Article 19 which says as follows:

“1. Member States shall require IORPs registered or authorised in their territories to invest in accordance with the ‘prudent person’ rule and in particular in accordance with the following rules:

(a) the assets shall be invested in the best long-term interests of members and beneficiaries as a whole. In the case of a potential conflict of interest, an IORP, or the entity which manages its portfolio, shall ensure that the investment is made in the sole interest of members and beneficiaries;

(b) within the prudent person rule, Member States shall allow IORPs to take into account the potential long-term impact of investment decisions on environmental, social, and governance factors;

(c) the assets shall be invested in such a manner as to ensure the security, quality, liquidity and profitability of the portfolio as a whole;

(d) the assets shall be predominantly invested on regulated markets. Investment in assets which are not admitted to trading on a regulated financial market must in any event be kept to prudent levels;

(e) investment in derivative instruments shall be possible insofar as such instruments contribute to a reduction in investment risks or facilitate efficient portfolio management. They must be valued on a prudent basis, taking into account the underlying asset, and included in the valuation of an IORP’s assets. IORPs shall also avoid excessive risk exposure to a single counterparty and to other derivative operations;

(f) the assets shall be properly diversified in such a way as to avoid excessive. reliance on any particular asset, issuer or group of undertakings and accumulations of risk in the portfolio as a whole

Investments in assets issued by the same issuer or by issuers belonging to the same group shall not expose an IORP to excessive risk concentration;

(g) investment in the sponsoring undertaking shall be no more than 5 % of the portfolio as a whole and, when the sponsoring undertaking belongs to a group, investment in the undertakings belonging to the same group as the sponsoring undertaking shall not be more than 10 % of the portfolio.

Where an IORP is sponsored by a number of undertakings, investment in those sponsoring undertakings shall be made prudently, taking into account the need for proper diversification.

Member States may decide not to apply the requirements referred to in points (f) and (g) to investment in government bonds.”

(emphasis added)

1.3 It is, perhaps, worth noting that the “prudent person” rule has not, as such, been transposed into UK domestic legislation. Instead, the transposition presumption is that, as a matter of UK trust law, the prudent person rule already applies. There is more detail on how the legal framework for investing pension scheme assets fits together in the article referred to in the footnote below[29].

The Occupational Pension Schemes (Investment) Regulations, 2005, Regulation 4 (as amended)

2.1 The Occupational Pension Schemes (Investment) Regulations 2005 (the “Investment Regulations”), Regulation 4(1) transposes, into UK domestic law, part of the requirements of Article 19 of the IORP II Directive referred to in 1 above.

2.2 Regulation 4(1) provides as follows:

“(1) The trustees of a trust scheme must exercise their powers of investment, and any fund manager to whom any discretion has been delegated under section 34 of the 1995 Act (power of investment and delegation) must exercise the discretion, in accordance with the following provisions of this regulation.

(2) The assets must be invested—

(a) in the best interests of members and beneficiaries; and

(b) in the case of a potential conflict of interest, in the sole interest of members and beneficiaries.

(3) The powers of investment, or the discretion, must be exercised in a manner calculated to ensure the security, quality, liquidity and profitability of the portfolio as a whole.

(4) Assets held to cover the scheme’s technical provisions must also be invested in a manner appropriate[30] to the nature and duration of the expected future retirement benefits payable under the scheme.

(5) The assets of the scheme must consist predominantly of investments admitted to trading on regulated markets.

(6) Investment in assets which are not admitted to trading on such markets must in any event be kept to a prudent level.

(7) The assets of the scheme must be properly diversified in such a way as to avoid excessive reliance on any particular asset, issuer or group of undertakings and so as to avoid accumulations of risk in the portfolio as a whole. Investments in assets issued by the same issuer or by issuers belonging to the same group must not expose the scheme to excessive risk concentration.[31]

(8) Investment in derivative instruments may be made only in so far as they—

(a) contribute to a reduction of risks[32]; or

(b) facilitate efficient portfolio management (including the reduction of cost or the generation of additional capital or income with an acceptable level of risk),

and any such investment must be made and managed so as to avoid excessive risk exposure to a single counterparty and to other derivative operations.

(9) [Relates to treatment of qualifying insurance policies and look through provision for collective investment scheme investments].

(10) [Relates to deemed satisfaction of diversification requirement for qualifying insurance policies].

(11) In this regulation—

“beneficiary”, in relation to a scheme, means a person, other than a member of the scheme, who is entitled to the payment of benefits under the scheme:

[definitions of “derivative instrument”, “regulated market” and “technical provisions” omitted].

(emphasis added)

2.3 It is difficult to see how investment in bonds with a negative nominal yield contributes to the profitability of the portfolio as a whole (and the same may also be argued in relation to bonds which have a negative yield in real terms).

Conflict between DB Scheme trustee investment duties and the proposed LTO

3.1 Prima facie there appears to be some conflict between:

the duties of DB Scheme trustees when investing scheme assets as outlined in the legislation referred to in Section 1 and Section 2 above, and
the proposed LTO.

3.2 This is not addressed in Clause 123[33] of the Pension Schemes Bill 2020 which provides as follows:

“123 Funding of defined benefit schemes

(1) Schedule 10 contains—

(a) in Part 1, amendments of Part 3 of the Pensions Act 2004 (scheme funding), and

(b) in Part 2, minor and consequential amendments relating to the amendments mentioned in paragraph (a)

(2) In exercising any powers to make regulations, or otherwise to prescribe any matter or principle, under Part 3 of the Pensions Act 2004 (scheme funding) as amended by Schedule 10, the Secretary of State must ensure that—

(a) schemes that are expected to remain open to new members, either indefinitely or for a significant period of time, are treated differently from schemes that are not;

(b) scheme liquidity is balanced with scheme maturity;

(c) there is a correlation between appropriate investment risk and scheme maturity;

(d) affordability of contributions to employers is maintained;

(e) affordability of contributions to members is maintained;

(f) the closure of schemes that are expected to remain open to new members, either indefinitely or for a significant period of time, is not accelerated; and

(g) trustees retain sufficient discretion to be able to comply with their duty to act in the best interests of their beneficiaries.”

3.3 Schedule 10 of the Pensions Scheme Bill 2020 makes provision for amendments to Part 3 of the Pensions Act 2004 to:

make provision for DB Schemes to have a funding and investment strategy, and
for that funding and investment strategy to be recorded in a statement of strategy.

3.4 That said, if compliance with the proposed DB funding code would lead to DB Scheme trustees being in breach of their investment duties, then it would follow that the DB Scheme trustees could not comply with the proposed DB funding code to that extent.

Part III: Reliance on the GAD modelling as the key foundation
for proposed quantified LTO

Preliminary
The Consultation Document places very considerable reliance on the GAD Modelling Document.
That modelling is based on an economic scenario generator (“ESG”).
The ESG, in turn, is being used to model future outturns.
The problem is that the modelling of future outturns in this context is substantially dependent on:

4.1 the base line calibration of the model (see Section B and Section C below),

4.2 future investment returns by reference to various asset classes, and

4.3 future inflation levels.

Note: The conjoined probabilities of:

(a) employer insolvency, and

(b) funding level of the DB Scheme relative to buy-out cost at the time in question,

are not modelled[34].

With the likely exception of investments already held in gilts, the problem is that over the time horizons of 10, 15, 20 or 25 (or longer) periods, no-one can, in fact predict any of those matters with any degree of accuracy.
Is it valid to calibrate the Economic Scenario Generator used in the GAD Modelling Document so that it provides a projected yield path very close to that implied by the market as at 31^st March, 2018 (the effective date of the modelling)?
A question for consideration is whether yields implied by the market on gilts (whether fixed interest or index linked gilts) are a valid basis for forecasting (or making assumptions as to) future gilt yields and future inflation.
Consider the following:

“It is easy to see a positive contemporaneous relation between interest rates and preceding long-term inflation rates for much of the time – especially the most recent half century – but there appears to be practically no relation between long-term interest rates and future long-term inflation.

It is the future inflation that ought to matter more if investors successfully priced long-term bonds to protect their real returns from inflation over the future life of the bond they are investing in, just the opposite of what we see.”

(emphasis added)

Source: Irrational Exuberance, Chapter 2, page 13, Third Edition, published by Princeton University Press 2015. Author: Robert J. Shiller.

On the same page of the same book (and as already noted earlier in this document) the author observes:

“Although interest rates must have some effect on the market, stock prices do not show any simple or consistent relation with interest rates.”

(emphasis added)

The author is writing by reference to US market conditions. But there seems no reason to doubt the validity of conclusions drawn from the data observed as being equally applicable to UK gilt yields.

Note: A similar point is made in Part II, Section B5.1 in Baroness Altmann’s evidence to the House of Commons Treasury Committee where she writes:

“Indeed, traditional estimates suggest that a 1 percentage point fall in gilt yields leads to approximately a 20% rise in pension liabilities, but only a 6-10% rise in typical pension fund asset values.”

However, if there is evidence to the contrary, it would be helpful if this could be identified and published in the response to the consultation exercise.
Equally, if there is no such evidence, it would be helpful to have an explanation as to why the modelling used as the basis for the preliminary conclusions in relation to the LTO set out in paragraph 638 and 639 of the Consultation Document has been calibrated in the way in question.
It would be particularly helpful if the Pensions Regulator could publish research showing:

7.1 the extent to which valuation assumptions as to future returns for particular asset classes in the valuations already submitted to the Pensions Regulator,

7.2 demonstrated any predictive ability, when evaluated by reference to actual outturns for the asset class in question, over 5, 10, and 15 year time horizons by reference to the actual returns on the assets classes over those 4 periods (where those periods have already elapsed).

The Pensions Regulator should hold valuations for DB Schemes starting in 2006 (so more or less 15 years’ worth of valuations). It should, therefore, be possible to see what predictive ability the investment return assumptions have for different asset classes relative to actual investment returns over the periods in question.
It will also be helpful to identify the expected margin of error (insofar as that is capable of being predicted – is it 25%, 50%, 75% or 100% – or is it simply unknown because investment returns over long periods of time are simply not capable of being predicted or forecast with any meaningful degree of accuracy?).
For example, would the analysis referred to in 8 above reveal:

10.1 over-optimism (or over estimation) of returns in times of buoyant investment markets with higher recent returns, and

10.2 over-pessimism (or under estimates) in times of depressed investment markets with lower recent returns.

There is a related question as to the extent to which the assumptions have been affected by cognitive biases (see Section C4 below).
The gilts yield assumption in the GAD Modelling Document
Introduction

1.1 The modelling in the GAD Modelling Document assumes that discount rates will be based on a gilt yield plus an appropriate margin[35].

1.2 The GAD Modelling Document says as follows in paragraphs 2.11, 2.12 and 2.13, as follows:

“2.11 Outcomes are simulated stochastically, by running through 1,000 future economic scenarios from an economic scenario generator (ESG) provided by a market leading scenario supplier. The ESG produces distributions of possible annual asset returns and economic factors such as price inflation and interest rates. These scenarios are used to project scheme asset and liability values on an annual basis, which enables calculation of the scheme funding level each year.

2.12 Current market conditions imply UK interest rates will stay low for decades. We have used an ESG[36] that has been calibrated to be consistent with market conditions as at 31 March 2018 as shown in Chart 3. The dashed lines show the market implied fixed and index-linked gilt yields at various time points, and the solid lines show the ESG assumed yields. This shows the ESG calibration provides a projected yield path very close to that implied by the market.

2.13 Some commentators suggest that market implied yields are reflecting distortions in the supply and demand for gilts, which will not be sustained. We have therefore also tested an alternative approach which assumes real gilt yields converge towards higher levels, than the market implies, in the future. Appendix C provides further details of the analysis on an alternative set of economic scenarios, which suggests based on the LD basis and investment strategies modelled scheme outcomes are not materially altered as a result. This is largely due to the high levels of interest rate and inflation hedging included in the investment strategies modelled.”

(emphasis added)

Comment: It would be helpful to have an understanding of the outturns under the alternative scenario if either no (or reduced) interest rate and inflation rate hedging were included.

1.3 As noted in Part II, Section D, it is for consideration whether the number of investment strategies modelled are ones which DB Scheme trustees could properly implement without breaching the duties imposed on them when investing DB Scheme assets, whether through:

(a) the breach of the prudent person rule (investing with a view to losing money in real terms),

(b) investing without seeking to invest the portfolio profitably (ie not losing money in real terms),

(c) avoiding excessive issuer concentration risk (for example if the issuer or an associate of the issuer is manipulating the market in which the issuer’s bonds are traded), and

(d) whether the three times leveraged LDI strategy is complying with the restrictions on the use of derivatives in Regulation 4(1)(8) of the Investment Regulations (as the strategy needs to contribute to a reduction in risk- in some scenarios (eg rapid rise in long term interest rates), it appears to amplify risk[37].

1.4 The contrary argument is that the DB Scheme trustee:

(a) who seeks to invest the scheme assets in a way

which matches

(b) the way the present value of the future pension payment obligations are measured periodically,

is seeking to maximise the likelihood that member benefits will be paid in full in the event of the sponsoring employer becoming insolvent.

1.5 Of course, investing the scheme assets so that their value moves in line with the way in which the present amount, as at any valuation date, of the scheme’s future pension payment obligations is measured misses the point. The future pension payment obligations depend on having enough money to make the payments as and when they fall due.

1.6 If the objective of the investment approach is to lose money in real terms on the investments, that does not appear to be the proper application of the prudent person rule but rather a case of acting with reckless prudence.

1.7 Rather than investing with a view to reducing risk on reliance on the sponsoring employer, it becomes a case of investing with a view to:

increasing the amounts that have to be contributed to the scheme by the sponsoring employer, and
thereby reducing the amount the sponsoring employer can invest in growing its business, and
thereby reducing the strength of the employer covenant.

Predicting future investment returns, forecasting future investment returns or projecting future investment returns by reference to assumptions made over periods of ten or more years

2.1 As noted in Section B2, if long term yields implied by the market on gilts have no predictive power as to future inflation or as to future interest rates, this begs the question as to the use of the model output.

2.2 Similarly, as noted in Section B2 above, if there is no simple or consistent relationship between interest rates and expected returns on shares (or for that matter, other asset classes), why is a fixed margin over a gilt yield used to predict the future returns on the asset class in question.

2.3 Effective feedback is one of the methods to improve performance.As mentioned in Section B7, it would be helpful to publish information about the predictive power of asset class returns made in past valuation assumptions compared to actual outturn.

2.4 This approach of “keeping the score” is one of the rules identified in Superforecasting: The Art and Science of Prediction by Philip E. Tetlock and Dan Gardner, (2015). There is more in this review to that book in Futurecasts Journal, April 2016, available at this link (accessed on 31^st August, 2020): http://www.futurecasts.com/J)%20Tetlock%20&%20Gardner,%20Superforecasting%20II.htm. The book review author’s world view is different and is of interest (and, perhaps, to be taken with a pinch of salt).

Is it prudent to conclude from current market conditions that UK interest rates will stay low for decades? Is this an example of one or more cognitive biases?

3.1 Section 2 above notes the uncertainties in relation to predicting the future and asks the questions whether a prudent approach has been used in selection of the assumptions.

3.2 I am using the word “prudent” in a different manner to the way the word “prudence” is used in the GAD Modelling Document at paragraph 5.5 which says:

“The assumptions adopted for the analysis, both in terms of benefit cashflows and economic changes, are intended to provide a reasonable projection of possible outcomes without material bias for prudence or optimism.

However, it should be recognised that different outcomes remain possible. The economic simulations underpinning the analysis are provided by a market leading scenario provider and calibrated closely to conditions and market implied interest rate paths as at 31 March 2018.

Many asset-liability models do not calibrate as tightly to market prices, though as we assume that the scheme is fully hedged in the period after reaching significant maturity, the sensitivity of the analysis to the scenario calibration is limited.”

3.3 It is unclear whether the rules identified for improving predictions referred to in Superforecasting: The Art and Science of Prediction (see 2.4 above) have been considered in the modelling approach.

Cognitive biases and selecting assumptions for modelling purposes

4.1 Decisions on selecting assumptions for modelling purposes can be clouded by one or more of the following cognitive biases:

recency[38],
availability[39],
anchoring[40], and
prospect theory[41].

4.2 The description of the assumptions used (see the GAD Modelling Document, paragraph 2.12 as set out in 1.2 above) suggests an impact of these cognitive biases.

Consider the results of the modelling if it had been done as at 31^st March, 1980

5.1 To tease out the cognitive bias issues, Chart 4 below illustrates inflation trends in the UK since 1970.

Chart 4[42]

5.2 If the same ESG had been used and the same modelling was being done with the ESG being calibrated to the conditions as at 31st March, 1980, the gilt yields would have been somewhat different.

5.3 In September 1981, the yield on UK 30 year gilt was 16.01%.

5.4 That would be the long term yield implied by the market. But that would be proved incorrect by a very large margin relative to rates two decades later.

5.5 This “thought experiment” is not suggesting that the modelling should start from the assumption that the position as at 31st March, 2018 is the same as the position as at 31st March, 1980.

5.6 Furthermore, it does not follow that the future will be like the past.

5.7 But given the uncertainty as to how the quantitative easing exercise will end, it does raise the question:

(a) whether the output of the modelling is being accorded a degree of authority by the Pensions Regulator in its preliminary conclusions in paragraphs 638 and 639 of the Consultation Document,

(b) that is not justified by the limitations of the modelling process.

A look at long gilt yields and the rate of CPI inflation between March 2018 and March 2020

6.1 Table 1 below compares the position as at March 2018, March 2019 and March 2020 as between the long gilt yield for that month and the rate of CPI inflation for the same month (and also includes a column showing the extent which a return of gilts + 0.5% would have exceeded the rate of inflation).

Table 1

Month and Year	Long gilt yield	Long gilt yield + 0.5%	CPI rate	Return of long gilt yield + 0.5% relative to CPI rate
March 2018	1.82%	2.32%	2.5%	(0.18%)
March 2019	1.65%	2.15%	1.9%	0.25%
March 2020	0.85%	1.35%	1.5%	(0.15%)

Note 1: The long gilt yield for the month in question is the highest of the 4 yields (short, medium, long and ultra-long) for that month[43].

Note 2: Office for National Statistics[44].

6.2 It can be observed from Table 1 that:

(a) the long gilt yield has more than halved over two years (ie not well predicted by the market), and

(b) the CPI inflation rate has reduced by 40% over the same two year period.

6.3 Interestingly, the aggregate of the arithmetic amounts payable from the example scheme modelled in the GAD Modelling Document (assuming no increases for inflation) would appear to be smaller[45] than the technical provision calculated as at 31st March, 2018. But that is a function of a discount rate which is lower than the inflation rate assumed in projecting the future benefit payments.

Models and predicting the future: Use of model outputs
Accurately predicting future investment returns over 5, 10, 15, 20 or longer periods of time is very difficult (if not impossible).
The information which is available to the person seeking to predict future returns comprises:

2.1 information about past returns, and

2.2 information about current political and economic factors.

Models may be helpful to process that information and generate a range of outturns and provide projections that may be helpful. But:

3.1 by definition they are simplifications of complex real-world systems,

3.2 suffer from a GIGO risk[46],

3.3 suffer from the risk of being affected by those cognitive biases referred to in C4 above.

Because the results of models provide a precise number[47] (including the assumptions made in relation to future events) it is not necessarily the case that the users of the model information understand:

4.1 the validity of the assumptions used in the model, and

4.2 the limitations of the model.

There is then the risk that such consumers jump to the conclusion that the projections made by the model provide a reasonable accurate prediction of the future.
At this point it is worth drawing out the difference in terminology between:

a projection, and
a prediction or forecast.

In this context, the GAD Modelling Document describes the purpose of the analysis produced as follows in paragraph 1.4:

“TPR is looking to implement Fast Track and Bespoke frameworks for assessing a scheme’s performance against the LTO and are interested in testing the impact on scheme funding and member benefits of different approaches to LTO guidance and the transition for schemes from the current regulatory approach to any new requirements.

To assist their consideration of the different possible approaches, TPR has asked the Government Actuary’s Department (GAD) to model a number of combinations of LTO objectives, including a range of LD funding levels, points of significant maturity (measured by duration), investment strategies, projection periods (post the significant maturity point) and market conditions.

The analysis produced enables comparisons to be made of the extent of future scheme self-sufficiency and gives indicative probabilities of reaching sufficient funding to buy-out with an insurance company or, in poor scenarios, the shortfall in benefits members may be exposed to.

This analysis will be used to assist TPR in assessing the appropriate LD funding level to be used in implementing the Fast Track framework.”

(emphasis added)

If the key assumptions used in the model are not robust (see discussion in Section B and Section C above), this undermines the reliability and utility of the “indicative probabilities”.
These issues are explored further in a paper “Economic Thought and Actuarial Practice” published on 11th June, 2019 by the Actuarial Research Centre[48].
The section on economic scenario generators is of particular interest in drawing out the extent to which decisions using such models can be relied on. For example:

“Model risk and uncertainty

The final area of interest on ESGs concerned model risk. One thing that emerged very clearly from the interviews with those involved in ESGs and modelling of this type was an understanding of the presence of model risk and a detailed and nuanced understanding of the limits of models. However, a common theme was that many other users of actuarial information were not aware of this and a lack of understanding in this context clearly has implications and is consistent with the concern arising about the public understanding of the difference between a forecast and a projection.

“Because there’s a degree of ignorance at board level or at the level of senior employees who simply can’t have the technical knowledge to understand whether or not the model is reliable or its results. The people who are involved in the everyday application of the model can’t necessarily explain what people at board level want to know. Of course, the board or other senior employees don’t want to admit ignorance and those using the model don’t want to admit that they are providing stuff that’s not useful. And you sort of have the possibility of a sort of Emperor has no clothes situation.”

Academic Economist

The organisational reality of how models are developed and built is again well-understood by those who build them, and this has implications for the application of model output and how it is used and interpreted.

“The larger the organisation the more you have to find ways to subdivide that responsibility for understanding what are the different bits of the model, and you do end up with models where no one individual understands everything that’s going on inside the model … And I would be surprised if anybody would make a claim that they did understand the whole of the models, they’re all that complicated. So, while you might say it’s undesirable that people make decisions based on models that they don’t really understand, actually for an organisation beyond a certain size that’s inevitable. So, if you’re going to say, well that shouldn’t happen at all, then you’re going to say, organisations should not get bigger than this size where somebody can get their head around it.”

Former Partner in an actuarial consultancy

The last part of the story around model risk is the fact that there is a lack of honest discussion and communication about the limits of models. This is particularly true when it comes to the difference between risk and uncertainty.

“What that really comes down to on the side of the model is model risk and the fact if you speak to most people that are working with models they say, “Well, it’s OK. But…” And they’ve always got caveats for everything, because they understand that this is a very useful tool, but its limits are manifest and there are plenty of them. And that conversation never goes out all the way. Of all the problems that we see, how much of it is driven by a lack of honesty about model risk and the inability of models to deal with uncertainty? Because models are really good at probabilistic risk. But actually, they’re not very good at uncertainty, because nothing can really model that.”

Academic Economist

One thing to note in all of the discussion is the absence of any mention of the baseline risk when using models; that is to say we do not know the status quo prevailing at a point in time as it is that point that any ESG models will expand from i.e. projecting forward in 2007 vs 2008 would generate hugely different outcomes.”[49]

(emphasis added)

The GAD Modelling Document perfectly properly identifies the calibration of the model to yields which are implied by the market as at 31st March, 2018[50]. As noted in Table 1 above, there has been a more than 50% change in the long gilt yields from 31st March, 2018 and 2019. It would be helpful to have an understanding as to how that impacts on the base line risk referred to in 10 above.
To be fair, it should be noted that paragraph 638 of the Consultation Document does open with the words “While the GAD modelling has its limitations” but it then goes on to say “it nevertheless shows that an LTO can be set so that there is a higher likelihood of a typical mature closed scheme funded on a low dependency level and invested on a low risk basis being able to survive on its own with minimal reliance on employer support”.
In other words, it would appear that the GAD modelling is the most important factor in arriving at the preliminary conclusions reached in paragraph 638 of the Consultation Document.
There is, however, no discussion of the limitations of the model or any reasoning as to why reliance on the model for the conclusion was justified (or any discussion of baseline risk).
It is, perhaps, appropriate in this context to consider the following:

“The UK 2004 Pensions Act requires these schemes[51] to compute a ‘technical valuation’ of their liabilities. This requires a discounted cash flow calculation using projections of prices, earnings and investment returns over the life of the scheme, which by its nature will exceed fifty years. Trustees of the scheme must compare this number with the current assets of the scheme and take steps to limit any deficit.

Of course, no-one has any idea what prices, earnings and investment returns will be in fifty years’ time. Unavoidably ignorant of all but a few of the numbers they need to complete their spreadsheet, the actuaries who advise these schemes invent all the numbers (technically the responsibility for verifying their assumptions lies with the scheme trustees, who have even less idea what the relevant numbers might be).

…The regulatory regime seeks to reduce risk – in a world of radical uncertainty risk can never be eliminated – by prescribing a reference narrative so demanding and financially unattractive that no-one will sensibly aspire to it. A combination of well-meaning but misguided regulation and the misuse of models has materially reduced the prospects of a secure retirement for a majority of the British population.”

(emphasis added)

Source: Radical Uncertainty: Decision-making for an Unknowable Future, Chapter 17, pages 312 and 313, 2020 The Bridge Street Press, Authors: John Kay and Mervyn King.

More generally, in a time of rigged gilt markets (via quantitive easing), a regulatory approach which promotes buying gilts or other bonds to achieve the LTO (and the associated post LTO investment strategy (at negative real rates of return) heavily based on modelling:

“to protect member benefits, it is important to maintain an investment strategy that is highly resilient to risk”[52]

appears to have some analogy with this year’s A level results and the pursuit of the target of no grade inflation via the use of algorithms (or models) in not looking at the bigger picture.

Conclusion
The modelling outlined in the GAD Modelling Document has been carried out using one particular ESG calibrated in a particular manner.

2 From that one model with that particular calibration, the Pensions Regulator has reached the preliminary conclusions in paragraph 638 and 639 of the Consultation Document.

Given the amounts of money involved in going down the route identified in the preliminary conclusions, it would be helpful to have:

3.1 an understanding of the Pension Regulator’s decision taking process in relation to the use of the modelling results and its degree of understanding of the limitations of the modelling results, and

3.2 the extent to which there was robust challenge of the modelling.

In particular, it would be interesting to know what the results would look like if the central expectation that interest rates would remain low for decades [53] were changed to see a substantial increase in inflation rates over the next one or two decades (ie the point that quantitative easing will or may be assumed to have ceased and the consequences that have flowed from that).
If I were a decision taker at the Pensions Regulator, I would have “kicked the tyres” in this manner. Perhaps, this has been done. If so, it would be helpful to publish the additional information.
Unfortunately, there are are times when cognitive biases lead to models being used to support largely predetermined conclusions.

Part IV: Whether a quantified LTO will encourage, or further encourage, procyclicality in the investment of DB Scheme assets and the creation/enlargement of systemic risk in DB Schemes

Introduction

In this part I raise two issues:

is the proposed LTO (and the journey plan to the LTO) the encouragement (or further encouragement of procyclical behaviour)?, and
is the encouragement of the associated “low risk “ investment strategies increasing systemic risk in relation to DB Schemes (at the encouragement of the Pensions Regulator). In other words, a case of putting all of your eggs in one basket.
Procyclical behaviour
Is regulation which encourages pro-cyclical behaviour a good idea?

1.1 The Bank of England is using quantitative easing as a tool to stimulate economic growth.

1.2 However, the valuation methodology used in the GAD Modelling Document derives a discount rate/investment return on the assets, by reference to the expected return on the asset class in questions over the gilt yield.

1.3 Using gilt yields as the reference benchmark leads to behavioural change in the way in which DB Schemes invest their assets so as to become better aligned with the valuation mythology. That, in turn, leads to pension funds buying more and more bonds at increasingly low yields.

1.4 The proposed LTO (including the journey plan to attaining the LTO) further encourages that approach.

1.5 It follows that the LTO proposal, along with the various risk management proposals, in the Consultation Document are directly encouraging pro-cyclical behaviour rather than the opposite.

1.6 An alternative approach would be for the proposed replacement DB Funding Code to encourage DB Schemes:

(a) actively to avoid investments which are pro-cyclical, and

(b) to take the opportunity to reduce their holdings of gilts and bonds at a time of quantitative easing and to invest in other asset classes (as envisaged by the Bank of England (see Part II, Section B3 above).

1.7 For a further discussion on procyclical and structural trends in investment allocation by pension funds, see the Bank of England Discussion Paper on this topic available at this link: https://www.bankofengland.co.uk/-/media/boe/files/paper/2014/procyclicality-and-structural-trends-in-investment.

1.8 This Bank of England Discussion Paper pre-dates the more recent quantitative easing measures taken by the Bank of England and, by definition, does not consider the implications of proposed LTO.

Does a prudent long term investor bet against the house in a rigged bond market?

2.1 There is an associated question. Does a prudent long term investor invest in a bond market which he knows to be rigged, so that the price of the investment is artificially higher than it should be?

2.2 Prima facie, this would not appear to be prudent.

2.3 Whether you are running a fish and chip shop or a multi-billion pound business, you do not engage in that business activity with the ex ante expectation of making a loss (and if you carry on making losses, you go out of business).

Does the LTO (and associated quantification of assumptions in determining the LTO) create/enlarge systemic risk in relation to DB Schemes?
The GAD Modelling Document notes at paragraph 5.7 that:

“The modelling covers 1,000 possible scenarios and the outputs presented illustrate the probability of certain outcomes based on these scenarios. In practice all schemes will face the same economic conditions at the same time and there is therefore risks from accumulations of risk if all scheme adopt similar strategies.”

(emphasis added)

What can be observed from the LTO proposed in the preliminary conclusion in the Consultation Document at paragraph 638 and 639 is to encourage DB Schemes to adopt a particular investment strategy (or group of investment strategies) similar to the types identified in the GAD Modelling Document.
This is very much a case of putting all of your eggs in the same basket.
It would be helpful to have an understanding of the analysis done by the Pensions Regulator as to whether the proposed LTO leads to the creation/enlargement of systemic risk in relation to DB Schemes.
For example, if there were a very substantial increase in interest rates, could that result in DB Schemes which had used an interest rate hedging strategy running out of eligible collateral to post (and having to sell other assets in adverse market conditions).
There is a related point insofar as eligible collateral is provided via the repo market, with the repos being rolled over every 3 or 6 months. Is that borrowing which is not temporary and for the purpose of providing liquidity.
Does that infringe the restrictions on borrowing by pension funds contained in the Occupational Pension Schemes (Investment) Regulations 2005, Regulation 5 which prohibits borrowing by schemes except for:

“borrowing made only for the purpose of providing liquidity for the scheme and on a temporary basis.”

This reflects Article 19(3) of the IORP II Directive:

“3, The home Member State shall prohibit IORPs from borrowing or acting as a guarantor on behalf of third parties. However, Member States may authorise IORPs to carry out some borrowing only for liquidity purposes and on a temporary basis.”

The DB Pension Scheme Leverage and Liquidity Survey prepared for the Pensions Regulator by OMB Research[54] identified the following:

“1.1.3 Investments involving leverage

Approaching half (45%) of all schemes had increased their use of leverage over the last five years; these schemes accounted for 58% of scheme assets.
A quarter (23%) of all schemes had increased their use of leverage in the last 12 months; these schemes accounted for 34% of scheme assets.
The notional principal of schemes’ leveraged investments totalled £498.5 billion; interest rate swaps were held by 62% of schemes and accounted for 43% of all leveraged investments.
The maximum permitted level of leverage ranged from 1x to 7x.
More than half of all schemes held all of their LDI assets either in segregated accounts (52%) or directly (4%). Just over a quarter (28%) of all schemes held all of their LDI assets in pooled accounts. A number of schemes (10%) held their LDI assets through a combination of arrangements.
Schemes held a total of £59.1 billion of outstanding gilt repurchase agreements and £6.9 billion of outstanding gilt reverse repurchase agreements.

1.1.4 Collateral monitoring

Basis points (bps) to exhaustion was the most commonly used method to estimate potential collateral needs under market stress; 55% of schemes used interest rate bps to exhaustion and 47% used inflation rate bps to exhaustion.
Half (53%) of schemes maintained a collateral ladder; together these schemes held £407.6 billion of assets.
A third (32%) of all assets were held in schemes where the trustees fully controlled the collateral management process.
Schemes where full control over the LDI assets, the collateral pool and additional scheme assets had been granted to the LDI managers (for the purposes of collateral management) accounted for 12% of all scheme assets.”

It is, perhaps, worth pondering whether, if there were to be a substantial and rapid rise in interest rates, a DB Scheme with substantial leverage (a 3 x LDI investment strategy?) could end up in financial ruin (subject to support from the sponsoring employer).
Conclusions
Going down the quantification and prescription route in relation to valuation assumptions and investment strategy will make some regulatory aspects easier.
It enables a “measured” comparison to be undertaken between different DB Schemes.
If what is being measured does not meet the regulatory yardstick, then the Regulator can specify action to be taken.
But the substitution of what some might, perhaps, unkindly call box ticking for judgment:

4.1 based on the proposed LTO (and the associated journey plan) leads to an extremely large cost (which is considered in Part V below), and

4.2 raises the question whether this provides value for money for the result achieved.

Part V: Cost and value for money relative to additional protection conferred in requiring DB Schemes to achieve a quantified LTO by the time they are substantially mature (as defined in the Consultation Document)

A reminder of some numbers used in the Consultation Document to provide context
Using the numbers in the Consultation Document at paragraph 98, the following (all values estimated as at 31st March, 2019) can be observed or deduced:

Table 2

	Item	£
1.1	aggregate DB Scheme assets	£1,720 billion
1.2	aggregate DB Scheme technical provisions	£1,900 billion
1.3	aggregate technical provisions basis deficit	£180 billion[55]
1.4	aggregate buy-out cost	£2,640 billion
1.5	aggregate buy-out deficit	£920 billion

As at 31st March, 2019, the amount to achieve (using the numbers set out in Table 2 above) the LTO for all DB Schemes estimated as at that date would be calculated as follows:

2.1 93% of £2,640 billion = £2,455 billion

Minus

2.2 £1,900 billion

Equals

2.3 £555 billion.

Note: I have assumed that all DB Schemes would aim to be fully funded on a technical provisions basis in accordance with the recovery plans that they are required to put in place under Part 3 of the Pensions Act 2004 (but that requires a further £180 billion to be made up).

To be clear:

3.1 the Consultation Document is not suggesting that the LTO is to be achieved immediately. Instead, the preliminary conclusion is that it should be achieved when the scheme is significantly mature (ie, with a duration, as that term is defined in the Consultation Document Glossary, of 14 years).

Note 1: It seems that the duration measurement of 14 years is highly sensitive to the interest rate levels. In other words, if interest rates are very low, the time before a DB Scheme would have a duration of 14 years would be substantially longer than if interest rates were substantially higher.

Note 2: However, if the DB Scheme’s assets have not moved in line with the increase in interest rates, there is additional cost on the employer to make good the shortfall if the LTO is to be attained earlier.

3.2 The GAD Modelling Document at Section 5.1 says:

“The GAD analysis has been carried out for TPR to help illustrate the relative levels of risk to member benefits and the PPF as a result of setting a long-term funding target at a variety of different levels. There is no modelling of sponsors so there is no allowance for insolvency or unpaid contributions in the period before scheme’s reach significant maturity”.

(emphasis added)

However, it is clear (and unsurprising) from the Consultation Document that to achieve the self-sufficiency funding level (ie. the LTO), requires the gap between:

4.1 93% of the buy-out cost (before expenses)[56], and

4.2 the amount of the technical provisions,

to be bridged by:

(a) additional contributions, or

(b) additional investment return (in excess of gilts plus 0.05%), or

As set out in 2 above, that amount as at 31st March, 2019 is estimated to be £555 billion.
Who bear the cost of the additional contributions?
Employers

That the employers bear the costs flowing from the additional contributions needed to attain the LTO may be viewed as self-evident.

Tax payers

2.1 But, where the employer is tax paying, there is also a cost to the tax payer (because of the tax deduction given in respect of the employer contributions).

2.2 For example, if the employer contributes £100, the cost to the tax payer can be assessed as being in the region of £20.[57]

Other stakeholders

3.1 If the employer is non-tax paying (e.g. a charity such as an educational establishment) it is the beneficiaries of the charity who will bear that cost.[58]

3.2 In other word, in the case of a university, it is the students paying fees of at least £9,450 a year who will ultimately bear the cost of the proposed approach to improving protection for member benefit.

Affordability and value for money: To gold plate or not to gold plate, that is the question
Introduction

If my objective is to have a teaspoon to stir my cup of tea with, I may be able to afford a gold plated teaspoon, but an ordinary one is just as effective.

Proportionate approach

2.1 From a value for money perspective, to be fair to the employer, the tax payer and the other stakeholders, the cost of providing the additional level of protection proposed for member benefits should be proportionate to the objective.

2.2 It may be argued that the only Pension Regulator objectives relevant in this context are those set out in Section 5(1)(c) and (cza) of the Pensions Act 2004 which are set out, for ease of reference, below:

“5 Regulator’s objectives

(1) The main objectives of the Regulator in exercising its functions are—

(a) to protect the benefits under occupational pension schemes of, or in respect of, members of such schemes,

(b) [relates to personal pensions]

(c) to reduce the risk of situations arising which may lead to compensation being payable from the Pension Protection Fund (see Part 2),

(cza) in relation to the exercise of its functions under Part 3 only, to minimise any adverse impact on the sustainable growth of an employer,

(ca) [not relevant]

(d) [not relevant].”

(emphasis added)

2.3 It can then be argued that there is no additional requirement:

to be fair to the employer, the tax payer or other stakeholders, or
to consider value for money (the value for money point is one which the Pensions Regulator is rightly keen to follow up on in relation to DC Schemes).

2.4 Reference is made in the Consultation Document at paragraphs 653 to 658 to the affordability of the cost of getting to the LTO (and implementing the LTO itself).

2.5 However, if the amounts involved are unknown (or not yet published), it is not possible to establish whether the risk reduction intended by the proposals for scheme members is value for money.

A £100 billion to £120 billion cost as at 31^st March, 2019 to achieve the LTO?

3.1 I noted, at A5 above, the cost of achieving the LTO as at 31st March, 2019 for all DB Schemes (aggregated) was estimated to be £555 billion).

3.2 Allowing for additional investment return (and the payment of pensions over the estimated period to achieve the LTO by the time DB Schemes are substantially mature – ie. with a duration, as that term is used in the Consultation Document, of 14 years) would reduce the £555 billion to a lower number.

3.3 A blog post[59]published by Con Keating and Dr. Iain Clacher[60] attempts to estimate the cost, calculated as at 31st March, 2019, of all DB Schemes achieving the LTO as identified in paragraph 68 of the Consultation Document.

3.4 Based on the modelling undertaken (as referred to in that blog post), there is a wide range of results with the 5% – 95% range being £80 billion to £200 billion and with the most common outcomes lying in the range of £100 billion to £120 billion (in all cases as at 31^st March, 2019).

A cost to the tax payer of between £20 billion and £24 billion (present capital value as at 31^st March, 2019)?

4.1 Assuming:

all employers would obtain tax deductions for these additional contributions with a present capital value as at 31st March, 2019 estimated, based on most common outcomes of modelling referred to in 4 above, estimated to be between £100 billion and £120 billion, and
the average rate of corporation tax (including the effect of tax deductions against profits and the income tax payable by members of partnerships or limited liability partners who are DB Scheme employers) remained at 20%,

the present capital value of those tax deductions, as at 31st March, 2019 would be between £20 billion and £24 billion.

4.2 Assuming that the figures calculated are in broadly the correct ball park, this raises the question whether the objective of the exercise can be achieved in a lower cost manner providing better value for money.

Striking a balance

5.1 As always there is a balance to be struck in managing risk and uncertainty between:

(a) protecting against the adverse outturn (employer insolvency) occurring:

(i) at a time when the scheme is not funded to a self-sufficiency level, and

(ii) benefits are reduced potentially to the PPF level (as the scheme assets are taken over by the PPF which provides compensation to the scheme members in return), and

(b) the cost of protecting against that risk.

5.2 It may be helpful to put the amounts of the loss of benefits to individual members in the context of the level of protection provided by the Pension Protection Fund (the “PPF”).

Reminder of the compensation levels provided by the PPF in assessing the level of benefit loss sustained by members if the scheme employer becomes insolvent and the scheme benefits are replaced by compensation from the PPF[61]

6.1 As noted in the extract from the White Paper at Part I, Section B3, the level of compensation provided by the PPF is high. For ease of reference it may be useful to spell this out in a bit more detail.

6.2 If the member has attained normal pension age for PPF purposes[62]:

(a) there is no reduction in the amount of the member’s accrued pension, but

(b) if the member’s scheme pension had attached to it pension increases relating to pre 6th April, 1997 pensionable service, those guaranteed increases are not covered, and

(c) in respect of guaranteed pension increases attaching to his scheme pension attributable to pensionable service after 5th April, 1997, those increases are limited to increases in line with the increase in the Consumer Prices Index (the “CPI”) capped at 2.5%.

Note: The monetary loss here is in respect of pensions derived from pensionable service incurred from 6^th April, 1997 to 5^th April, 2005 where the cap on increases by reference to the rate of inflation becomes 2.5% in place of 5%. If the level of guaranteed increase provided under the member’s scheme was at a higher level, then the member would see that higher level capped at the increase in the CPI (in turn capped at 2.5%).

6.3 If the member has not attained normal pension age for PPF purposes in relation to his accrued pension, then:

(a) the member’s accrued pension is reduced by 10% (and was subject to an additional cap – see 6.4(a) below),

(b) the future increases for that pension are reduced in the same way as set out in 6.2 above, and

(c) if the member’s pension is not yet in payment, in summary, the member’s revaluation of deferred pension is limited to the increase in the CPI over the number of complete years in the deferral period to normal pension age (or, if lower, 5% a year over the same number of complete years for pensionable service before 6 April 2009 and 2.5% a year over the same number of complete years for pensionable service after 5^th April, 2009).

Note 1: Special rules apply where pensionable service ended after 31^st December, 1985 but before 1^st January, 1991.

Note 2: No revaluation applies where pensionable service ended before 1^st January, 1986.

6.4 In this context, it may be helpful to note the following:

(a) as a result of the High Court decision in Hughes v. the Pension Protection Fund[63] (22nd June, 2020) the compensation cap applicable to a member falling within 6.3 above, has been set aside on grounds that it is age discriminatory[64].

Note 1: The standard compensation cap for a member with a normal pension age of 65 was £41,461 from 1^st April, 2020.

Note 2: For members with 21 or more years pensionable service, the standard compensation cap is increased by 3% to the standard amount applicable to the member for each full year of pensionable service over 20 years to a maximum of double the standard compensation amount applicable to the member.

(b) there is an additional underpin for members in 6.2 and 6.3 in respect of the loss of future pension increases (and revaluation for deferred members) which follows from the ECJ decision in Hampshire v. the Pension Protection Fund[65].

Conclusions
If the ballpark calculations set out in Section C above are broadly correct:

1.1 the move to putting in place a LTO (and the associated journey plan) will have an overall cost in the region of between £100 billion and £120 billion (with the tax payer bearing between £20 billion and £24 billion of that cost).

1.2 the counter factuals of the level of likely reductions to members’ benefits before the LTO is reached and of not adopting the LTO have not be assessed.

1.3 no case has been considered (or made) that the proposed LTO (and the associated journey plan) along with the proposed quantification approach in the Consultation Document are provide value for money.

1.4 it appears more likely than not that it provides very poor value for money and would cause significant collateral damage.

I have not attempted to quantify the impact on the current active members of those DB schemes which are open to future accrual (or, in some cases, to new active members as well as future accrual).
But, it would seem likely that:

(a) future accrual on a DB basis would stop, or

(b) there would be a further reduction in the level of DB benefits.

A possible alternative approach is considered in Part IV below.

Part VI An alternative approach for consideration

An alternative approach?
Preliminary

1.1 To be fair to the Pensions Regulator, the alternative approach which is suggested in this Part, is not one within which it is in the powers of the Pensions Regulator to implement.

1.2 That said, it is to be hoped that the points made in this document as to the suggested alternative solution would be the subject of further discussion with the Department for Work and Pensions (and HM Treasury) so that the policy decision in the White Paper can be achieved, albeit in a different and, if it is submitted, more cost effective manner.

You don’t need to reserve for the cost of rebuilding your house in case it should burn down if you can insure it

2.1 The analogy to the proposed alternative approach outlined below is that you do not hold a reserve equal to the rebuilding cost of your house in your savings (assuming you could afford to do so).

2.2 Instead, you pay an annual premium to insure against the risk in the amount of the rebuilding cost.

2.3 Because the risk of your house burning down is, in general, relatively low, the insurance market operates as a way of managing that risk in a more cost efficient manner.

2.4 So the question that arises is whether an insurance based (or mutual support fund) solution could be put in place at a substantially cheaper cost.

Outline of a mutual support fund solution
The PPF already exists. To date:

1.1 it has been successful in providing compensation for members of those schemes whose assets are taken over by the PPF, and

1.2 it has been running a surplus[67].

There can be a separate debate as to whether the PPF levies are too high or not. However, it is an example of something corresponding to a mutual insurance scheme that has worked.
The alternative approach which I am suggesting:

3.1 is based on a mutual insurance scheme similar to the PPF which is compulsory (“PPF2”).

3.2 PPF2 is funded by annual premiums (or levies if preferred).

3.3 the cover provides for any shortfall between the scheme benefits and the PPF benefits.

3.4 the premiums would be paid:

as to a specified percentage by the sponsoring employer, and
as to the balance by the scheme members or by HM Treasury (or shared between them) reflecting the cost to the tax payer of the estimated £20 billion- £24 billion loss of tax revenue from tax deductions on additional employer contributions.

3.5 the amounts to be paid by the scheme members would be collected through the reduction in the future revaluation/indexation of the member pensions, but with a floor so that the member pension could never fall below:

(a) the in payment pension at time of employer insolvency, and

(b) for the pension not in payment, the accrued pension to date of employer insolvency[68]

3.6 it may be that PPF2 could be managed by the PPF (but it would not be legally part of the PPF).

3.7 it could have the same safety valves as the PPF to reduce future revaluation and future increases and compensation levels, if necessary, to “balance the books”.

3.8 the alternative approach to that in 3.7 above would be for there to be a stop loss reinsurance provided by the Government (recognising that the Government has considerable “skin in the game” via the tax relief on employer contributions).

[1] Cm 9591, at page 5. Published in March 2019 and available here: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/693655/protecting-defined-benefit-pension-schemes.pdf

[2] The White Paper at page 5

[3] Philip.f.bennett@durham.ac.uk

[4] “Modelling the Long-Term Funding Objective: A likely outcome for different approaches” as supplemented by the information contained in the additional document dated 11th August, 2020.

[5] https://www.bankofengland.co.uk/monetary-policy/quantitative-easing

[6] On 20^th May, 2020 the Debt Management Office issued £3.8 billion of three year 0.75% gilts with an average redemption yield of minus 0.003% pa: see https://www.dmo.gov.uk/media/16554/200520conventional.pdf.

[7] https://www.ecb.europa.eu/stats/financial_markets_and_interest_rates/euro_area_yield_curves/html/index.en.html

[8] Information obtained from this link, accessed on 24th August, 2020: https://www.bankofengland.co.uk/monetary-policy/quantitative-easing

[9] For some further analysis as to the effect of quantitive easing on employer capacity to invest see here: https://www.bankofengland.co.uk/-/media/boe/files/working-paper/2018/growing-pension-deficits-and-the-expenditure-decisions-of-uk-companies.pdf .

[10] By Martin Daines, Michael A.S. Joyce and Matthew Tong, October 2012 at page 41, available here: https://www.bankofengland.co.uk/-/media/boe/files/working-paper/2012/qe-and-the-gilt-market-a-disaggregated-analysis.pdf

[11] Source: https://www.dmo.gov.uk/umbraco/surface/DataExport/GetDataExport?reportCode=D4H&exportFormatValue=doc&parameters=%26Start%20Month%3D7%26Start%20Year%3D2009%26End%20Month%3D7%26End%20Year%3D2009

[12] https://www.bankofengland.co.uk/monetary-policy/quantitative-easing

[13] Source: https://www.dmo.gov.uk/umbraco/surface/DataExport/GetDataExport?reportCode=D4H&exportFormatValue=doc&parameters=%26Start%20Month%3D7%26Start%20Year%3D2010%26End%20Month%3D7%26End%20Year%3D2010

[14] Source: https://www.dmo.gov.uk/umbraco/surface/DataExport/GetDataExport?reportCode=D4H&exportFormatValue=doc&parameters=%26Start%20Month%3D7%26Start%20Year%3D2020%26End%20Month%3D7%26End%20Year%3D2020

[15] More here: https://www.parliament.uk/business/committees/committees-a-z/commons-select/treasury-committee/inquiries1/parliament-2010/quantitative-easing/

[16] My Comment: But not all parts of the Bank of England appear to be of the same mind; see 3 above.

[17] Written evidence of Dr Ros Altmann (now Baroness Altmann) available at this link (pages 29 and 30): : https://publications.parliament.uk/pa/cm201213/cmselect/cmtreasy/writev/qe/qe.pdf

[18] Source: Irrational Exuberance, Chapter 2, page 13, Third Edition, published by Princeton University Press 2015. Author: Robert J. Shiller.

[19] I.e. the investor would receive interest payments on the due date and repayment of principal at maturity, in each case in full but taking no account of the effect of inflation (and, if a foreign investor, depreciation of sterling relative to the foreign investor’s currency of record).

[20] Extract from the Manifesto of the Soviet of Workers’ Delegates published in 1905.More at this link accessed on 24^th August, 2020:: https://www.cadtm.org/Russia-Origin-and-consequences-of-the-debt-repudiation-of-February-10-1918

[21] More here (accessed on 24^th August, 2020: https://www.economist.com/graphic-detail/2019/09/12/austrias-100-year-bond-has-delivered-stunning-returns

[22] On 20^th May, 2020 the Debt Management Office issued £3.8 billion of three year nominal gilts with a yield of minus 0.003% pa.

[23] Harold Wilson https://commonslibrary.parliament.uk/economy-business/economy-economy/pound-in-your-pocket-devaluation-50-years-on/.

[24] The powers to issue contribution notices and financial support directions contained in the Pensions Act 2004, Sections 38 to 57.

[25] If the investment income on the scheme assets plus contributions exceeds benefit outflow, the scheme is cash flow positive.

[26] The example scheme referred to in the GAD Modelling Document provides full RPI indexation (capped at 5% for post 5th April, 1997 pensionable service). See Appendix A of that document.

[27] See The White Paper at page 5.

[28] Directive (EU) 2016/2341 (replacing the IORP I Directive (Directive 2003/421/EC). In the IORP I Directive, Article 18 contained the investment provisions. The UK withdrew from the EU on 31^st January, 2020. But until changed by Parliament, EU law applicable to the UK as at 31^st January, 2020 remains part of UK law.

[29] Must an occupational pension scheme take into account ESG Factors even if there is a risk of financial detriment to the Pension Fund, Trust Law International, 219, Vol 32, p239 available at this link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3375750

[30] In compliance with the prudent person rule, there are a number of ways in which the assets covering the scheme’s technical provisions may be appropriately invested. The proposed quantification approach used in the Consultation Document seeks to constrain the range of those prudent approaches.

[31] Consider the position of a scheme which has an exposure to a single issuer in respect of, perhaps, 70% of its investments. Such a scenario is contemplated in the GAD Modelling Document at Table 5 on page 13 (the issuer is HM Treasury).

[32] Consider whether a three times leveraged LDI asset allocation envisaged by the GAD Modelling Document satisfies this requirement.

[33] As it stands after the third reading of the Pension Schemes Bill 2020 in the House of Lords.

[34] The GAD Modelling Document at paragraph 5.1 says “The GAD analysis has been carried out for TPR to help illustrate the relative levels of risk to member benefits and the PPF as a result of setting a long-term funding target at a variety of different levels. There is no modelling of sponsors so there is no allowance for insolvency or unpaid contributions in the period before schemes reach significant maturity”.

[35] See, for example, Tables 1, 2 and 3 in that document.

[36] Economic Scenario Generator (not to be confused with environmental, social and governance factors).

[37] More here: https://www.thepensionsregulator.gov.uk/-/media/thepensionsregulator/files/import/pdf/db-pension-scheme-leverage-and-liquidity-survey.ashx.

[38] More here: https://en.m.wikipedia.org/wiki/Recency_bias

[39] More here: https://www2.psych.ubc.ca/~schaller/Psyc590Readings/TverskyKahneman1974.pdf

[40] More here: https://en.m.wikipedia.org/wiki/Anchoring_(cognitive_bias)

[41] More here: https://en.m.wikipedia.org/wiki/Anchoring_(cognitive_bias)

[42] Accessed on 24th August, 2020 at this weblink: https://www.economicshelp.org/blog/5720/economics/inflation-stats-and-graphs/

[43] Details of source and an explanation of calculation are available at this link: https://www.dmo.gov.uk/data/gilt-market/ (accessed on 29th August, 2020)

[44] Available at this link https://www.ons.gov.uk/economy/inflationandpriceindices/timeseries/d7g7/mm23 (access on 29th August, 2020).

[45] I have not calculated what the expected RPI inflation rate would be as derived from the gilt curves but assume that it would be more than 1.82% a year (see GAD Modelling Document, Appendix A, page 33 for this actuarial assumption)

[46] I understand that, in the modelling world, this means Garbage In = Garbage Out.

[47] Or, in the case of stochastic model, a range of outturns which are accorded higher or lower levels of probability of occurrence.

[48] By Dr. Iain Clacher, available at this link https://www.actuaries.org.uk/system/files/field/document/Economic_Thought_and_Actuarial_Practice.pdf

[49] At page 12/13 : https://www.actuaries.org.uk/system/files/field/document/Economic_Thought_and_Actuarial_Practice_0.pdf

[50] The GAD Modelling Document, paragraph 2.12.

[51] i.e. DB Schemes.

[52] The Consultation Document, paragraph 639.

[53] See the GAD Modelling Document, paragraph 2.12

[54] Available at this link, access on 31st August, 2020: https://www.thepensionsregulator.gov.uk/-/media/thepensionsregulator/files/import/pdf/db-pension-scheme-leverage-and-liquidity-survey.ashx

[55] This figure is net of surpluses in some schemes (so the amount required to fill the deficits of those schemes in deficit will be correspondingly higher).

[56] 93% of the buy-out cost before expenses is the low dependency funding target referred to at paragraph 626 of the Consultation Document. It is adopted as the Pensions Regulator’s preliminary conclusion, at paragraph 638 of the Consultation Document to be reached by all DB Schemes at the time they are significantly mature (ie. duration, as at that time, is defined in the Consultation Document as 14 years).

[57] The current rate of corporation tax is 19%. However, where the employer is a professional services firm structured as a partnership, or a limited liability partnership, the members of that partnership are likely to be paying tax at 40% or 45% (hence the rounding up to 20%). If the current rate of Corporation Tax increases, then the cost to the tax payer would, likewise, increase.

[58] Because the services provided by the charity will have to reduce, all other things being equal.

[59] On 1^st September, 2020 available here: https://henrytapper.com/2020/09/01/bonfire-keating-and-clacher-conclude-their-articles-on-the-db-funding-code/

[60] Con Keating is Head of Research for the Brighton Rock Insurance Group. Dr. Iain Clacher, Leeds University Business School.

[61] For authority for the statements made in this Section, see the Pensions Act 2004, Schedule 7 (Pension Compensation Provisions) as amended and associated statutory instruments.

[62] Ie. the earliest date, ignoring special circumstances, at which the member may draw his benefits as of right unreduced.

[63] https://www.judiciary.uk/wp-content/uploads/2020/06/hughes-v-ppf-judgment-220620.pdf

[64] Leave to appeal this decision has been granted, but it seems unlikely that the appeal will be successful.

[65] http://curia.europa.eu/juris/document/document.jsf?text=&docid=205405&pageIndex=0&doclang=en That underpin would, however, be limited to 50% of the level of pension at any point in time which the member would have received from the scheme had the employer not become insolvent. Note that the PPF wishes to deal with this underpin by way of a capital value calculation and an appropriate increase rather than doing an annual test. That is the subject to a separate appeal from the High Court decision in Hughes v The Pension Protection Fund.

[66] http://curia.europa.eu/juris/document/document.jsf?text=&docid=221800&pageIndex=0&doclang=en&mode=req&dir=&occ=first&part=1&cid=41284

[67] For example see here: https://www.pensionsage.com/pa/PPF-could-share-expected-pension-surplus-with-members.php

[68] The same rules as in the PPF would apply for benefit uplifts in the run up to insolvency.

4 Responses to Philip Bennett’s DB funding code response

Bob Compton says:

September 4, 2020 at 7.32

Henry, Phillip’s consultation response is a superbly researched critique of the Pensions Regulators Scheme funding consultation. ARC’s consultation response left the majority of questions raised by TPR unanswered as we did not want to be giving credence to the thought process behind the questions, nor did we have the resources to do the research that Phillip has set out in his paper, Phillip should be congratulated for preparing his paper, which does a hatchet job on the proposition, and sets out very clearly the folly of following a gilt based investment strategy in the current climate.

Unfortunately I suspect it will be ignored by the Regulator, as it does not fit their current intended direction.

As a separate note I raised the issue of the impact on UK DB pensions schemes with Charlie Bean the then deputy governor of the bank of England on the introduction of QE. Charlie admitted in the public meeting that this was something that had not been considered, and that pensions were a long term issue whilst the BoE had an immediate issue to deal with. I pointed out the info on impact that Roz Altman alluded to with Charlie Bean, and it was ignored at that time. That was when £200m was to be issued, Today it is over 10 times that number, and the result has been the destruction of DB as a viable pensions delivery option for Employers.

If TPR’s proposal is implemented it will be the final nail in the coffin for DB schemes in the private sector.

Loading...

Ian Neale says:

September 4, 2020 at 7.32

Bravo!! I hope those at TPR charged with assessing responses to the consultation possess sufficient intellectual acuity to give due weight to your submission. If they are imbued with a proper sense of responsibility we might hope so, but it is always hard for humans to recognise when they are barking up the wrong tree, even when faced with such a comprehensive demolition of their position as delivered here.

Loading...

George Kirrin says:

September 5, 2020 at 7.32

I fear our confirmation biases are showing, Bob and Ian (and me).

Like you both, I enjoyed Professor Bennett’s demolition job, but only up to a point. I found the final section on alternative solutions – an extension of PPF “mutual support” or some form of stop-loss insurance – rather underwhelming, although Con Keating may have more to say on here about insurance solutions.

As for the PPF, I have previously questioned its low investment target of LIBOR + 1.8%, so I’d be expecting more from such a “mutual” solution, rather than higher levies all round.

I also did not detect (although I may have missed it) an alternative to gilts based discount rates. I’d like to hear more about “prudent expected returns”, and not just the usual warnings that expected returns are impossible to forecast because of future uncertainties. I believe more could be done about demystifying expected returns, if necessary taking them apart piece-by-piece in terms of expected interest payments, expected dividend payments, expected rental payments, currency effects, and other expected sources of return, whether contractual or simply reliant on a diversified portfolio of complementary assets.

I still believe (or at least hope) that a solution to funding can come from investment of finite resources, rather than just making seemingly infinite calls on scheme sponsors to keep diverting capital towards DB schemes and away from business maintenance and business growth. Some prudent re-risking instead of it always being about capital-intensive “de-risking”, please?

Loading...

henry tapper says:

September 6, 2020 at 7.32

Bob’s response which is now published has things to say about alternative discount rates

Loading...