Written evidence from John Ralfe LDI0071

Following my earlier written[1] and oral evidence[2], I would like to make some further comments to inform your evidence session with the Minister for Pensions and the Economic Secretary to the Treasury.

This includes my comments on whether the Pensions Regulator’s Funding Code for DB schemes [3], which will apply from October 2023, could “result in increased herding in pension scheme investments”.

The Committee also asked for comments on the Bank of England’s recommendations that “that TPR should take regulatory action, in coordination with the FCA and overseas regulators, to ensure LDI funds remain resilient and longer term”

I have read the WPSC evidence from Sarah Breeden, of the Bank of England, on its plans for “steady state” resilience for LDI funds. I have no specific expertise in this, but what we have heard so far seems entirely sensible.

My evidence also includes my recent letter to the Lords Committee explaining why it is wrong to suggest “accounting” is to blame for the LDI crisis, and why moving to discounting pension promises at the “expected return on assets” is wrong, in both theory and practice.

I would like the WPSC to consider this as the issues have also been raised in its earlier sessions.

1 Additional comments

What lessons can we start to draw about LDI?

1 There is a world of difference between Liability Driven Investment and “Leveraged LDI” – and it is Leveraged LDI, not vanilla LDI, which caused last year’s apparent meltdown.

LDI is just jargon for matching pension assets and liabilities, which Boots pioneered 20 years ago. As well as switching from equities to long-dated bonds, including index-linked, interest-rate swaps can also be used to improve matching, especially inflation matching, again as Boots pioneered 20 years ago.

Hedging pension liabilities reduces risk for scheme members, the sponsoring company, the PPF — which pays compensation if a sponsor goes bust — and the financial system as a whole.

But with “Leveraged LDI”, a pension scheme is effectively borrowing to buy assets which don’t match liabilities — equities, PE, hedge funds, property — a bet that their value will increase more than the value of liabilities. This is pure speculation.

The Bank of England Deputy Governor, Sam Woods, recently told the Treasury Select Committee: [4]

“ [T]here is a bit of having your cake and eating it: you keep the returns from the higher returning assets you have and you leverage for the gilts part that you need for matching purposes. Q311

Andrew Bailey, the Bank Governor, added:

“ [W]hat started as a means of managing asset liability positions became a means of actually increasing the return to the fund... that is the leverage point. Q314

2 This leverage was hidden from members, shareholders, and bondholders, because accounting requirements are poor — pension schemes and companies don’t have to disclose details of their Leveraged LDI.

And, at best, the different regulators had only partial information on Leveraged LDI for individual schemes, and for the whole financial system.

Poor understanding includes trustee boards. How many really understand the risks they are taking in opaque, complex and very expensive Leveraged LDI? This problem isn’t solved by more ‘professional’ trustees, as some people have suggested: we need “lay” trustees asking the “stupid” questions.

3 Despite some claims, pension schemes do not have to be leveraged.

Investment consultant Rod Goodyer told the WPSC that: [5]

“DB pension schemes are forced, by their very construction, to mismatch assets and liabilities.  There is therefore a tension between managing asset/liability mismatch (i.e. buying gilts to match liabilities) and generating the returns required in the funding plan to meet benefits in full.  This is the reason why leveraged LDI was invented.

The claim that Leveraged LDI “was invented” to make sure pensions can be paid is extraordinary. Schemes are not “forced” to mismatch assets and liabilities, they can simply choose to hold long-dated bonds.

To quote Andrew Bailey again:

“ The LDI model does not necessitate leverage; it is a way of managing the assets and liabilities of pension funds. But it has become more leveraged over time. Q279

What should happen now?

1 Some people — including me — have called for pension schemes to be banned from using Leveraged LDI. After all, they are prohibited by law from borrowing, [6] but can use derivatives, or Leveraged gilt funds, to get round this. And Leveraged LDI has led to “moral hazard” on a grand scale, with taxpayers underwriting the consequences.

But rather than an outright legal ban — not easy to draft or enforce without unintended consequences — we can achieve a “soft” ban, through tougher supervision by TPR, and more transparent accounting.

2 TPR should introduce regular, detailed scrutiny of all Leveraged LDI. Trustees would have to show they really understood what they were doing, and had systems in place to react quickly to liquidity problems, including a  credit line from the sponsor, which some companies have already put in place.

Introducing this extra scrutiny should be a priority for the Regulator’s new CEO, and could be incorporated into the new DB Funding Code.

3 The IASB and the UK FRC should require pension schemes and companies to clearly disclose any Leveraged LDI in their accounts, including “uncovered” or “naked” interest rate swaps, leveraged gilt repos, and leveraged gilt funds.

There is already anecdotal evidence that smaller schemes are cutting back on Leveraged Gilt Funds. Having to answer awkward questions from TPR, scheme members, shareholders and bondholders will see Leveraged LDI shrinking, perhaps dramatically.

2 Will the Pensions Regulator’s new DB Funding Code “result in increased herding in pension scheme investments” ?

1 TPR’s new DB Funding Code comes into force in October, and plenty has happened in the DB pensions world since the last Funding Code in 2014.

Virtually all schemes are closed to new accruals and therefore increasingly mature, and schemes have to a greater or lesser extent matched their liabilities and assets. Many have gone further and matched specific tranches of liabilities with an insurance buy-in, and some have been able to transfer all their liabilities in a full buy-out.

Reflecting these changes TPR says:

“It is expected that schemes will reduce reliance on their sponsoring employer as they reach maturity. It will require trustees to improve risk management and raise the bar for evidencing supportable risk taking.”

TPR sets out the principles:

27. For the purposes of the funding and investment strategy, trustees must assume that scheme assets will be invested in accordance with a low dependency investment allocation on and after the relevant date.
28. A low dependency investment allocation is an investment strategy under which:
• the cash flow from the investments is broadly matched with the payments under the scheme, and
• the value of the assets relative to the value of the scheme’s liabilities is highly resilient to short-term adverse changes in market conditions
29. Trustees must ensure that the risks in the assumed investment strategy are sufficiently prudent, so that no further contributions would be expected to be required from the employer if the scheme was fully funded.”

The  Funding Code gives a specific example:

“72. One example of meeting our expectation would be to have a low dependency investment allocation made up of the following asset classes:

• corporate bonds and government gilts: 85% (assuming appropriate maturities and durations for the bonds)
• growth assets: 15%

We consider a scheme adopting this type of approach as its low dependency investment allocation would be considered to meet the requirement to be broadly matched with the payment of pensions and other benefits under the scheme.”

2 The new DB Funding Code will certainly further encourage pension schemes to match assets and liabilities by holding bonds and gilts – the one-way direction of travel over the last 20 years.

But are the consequences of matching assets and liabilities good or bad?

The Committee has seen arm-waving from people who seem to say that matching assets and liabilities by holding bonds is bad, because it represents “herd behaviour”, but their precise objections are vague.

Although it isn’t spelled out, the claim of “herd behaviour” presumably means that hedging assets and liabilities weakens the strength of schemes, increasing the risk they will be unable to pay pensions.

But hedging reduces risk for scheme members, because the value of assets and liabilities move in tandem, whatever happens to long term interest rates,  inflation, or the value of financial assets.

It also reduces risk for the sponsoring company, the PPF and the financial system as a whole, because the level of “cross shareholdings” amongst companies, is reduced.

3 Those warning that the new DB Funding Code encourages “herd behaviour” may also mean that holding bonds and gilts sows the seeds of a future liquidity crisis, which we saw following the “mini” Budget’

But the LDI liquidity crisis was not caused by matching pension assets and liabilities. Rather it was caused by schemes buying Leveraged Gilt Funds, allowing them to then buy non-matching assets - equities, PE, HFs and property. As I have outlined, this opaque, complex and very expensive Leveraged LDI was hidden from scheme members, shareholders, bondholders and regulators, and not understood by pension trustees.

Again, as I have outlined, the way to address this is by tougher TPR scrutiny, and more transparent accounting, so shareholders and bond holders can see what is going on. It is not to “throw-the-baby-out” by trying to undermine the new DB Funding Code.

4 There have also been rumblings that if pension schemes buy bonds and sell equities – which has been happening for the last 20 years – that companies will be somehow starved of new equity capital.

But we should be clear that DB pension schemes are not a source of new equity capital, since company sponsors are simply using their own capital to buy shares in other companies, in a weird game of cross shareholding pass-the-parcel.

If companies want to continue holding “risk seeking” financial assets they can do this directly by issuing long term secured debt, and use the cash to buy equities, PE, HFs or property. If companies want to bet their own balance sheet, that’s between them and their shareholders and bondholders, but don’t do it with pension scheme money. 

3 Letter to Lord Hollick

Lord Hollick

Chair Industry and Regulators Committee

House of Lords

London SW1A 0PW                                         24th February 2023

Dear Lord Hollick

“The use of LDI strategies by pension funds”

Further to the private evidence I gave the Committee in November, I would like to comment on your February 7th letter [7] to Andrew Griffith at the Treasury, and Laura Trott at the DWP, and I have copied them on this letter. I understand you are attending the March WPSC meeting, when  the Ministers are giving evidence.

I would be grateful if you would publish this letter on the Committee’s website, as a part of the public debate.

1 Is pensions accounting really the villain in Leveraged LDI?

It is difficult to pin-down who is to blame for last year’s Leveraged Liability Driven Investment debacle, when the UK’s £1.5trn defined benefit pension system seemed to be in meltdown, triggered by September’s “mini”-Budget, and saved only by the Bank of England intervening to buy gilts.

The finger has been pointed at many culprits - ill-informed pension trustees not understanding what they were getting into, avaricious investment consultants lining their own pockets, unscrupulous companies betting with pension scheme money, and complacent regulators – the Bank of England, the Pensions Regulator and the FCA – all dozing at the wheel.

Given these possible culprits I am surprised by the Committee’s conclusion that the real villain in Leveraged LDI Cluedo, is Mr Grey the accountant.

The Committee’s first conclusion says forcefully that:

“The fundamental issue is that leveraged LDI has been created as a solution to an artificial problem created by accounting standards” , requiring pension liabilities to be “discounted at a low-risk market interest rate.“

Using the “low-risk interest rate” – the AA corporate bond rate – means that “when market interest rates fall, the discounted value of future liabilities is shown as having grown, with a corresponding increase in the scheme deficit”

Matching pension assets and liabilities through “LDI” - was “designed to avoid this accounting volatility by investing assets in bonds and gilts that had the same sensitivity to market interest rates as the discounted liabilities, so their values move up and down”

Instead of using AA corporate bond yields to calculate pension liabilities the Committee wants to use the “the expected long-run return” on assets held by a pension scheme. This would reduce the volatility of published pension liabilities, and, according to the Committee, remove, at source, the incentive for pension schemes to use LDI, preventing any repeat of the Leveraged LDI meltdown.

Since almost all schemes hold some equities, with a higher “expected return” than AA bonds, this change would not only reduce the volatility of liabilities, but immediately reduce liabilities, making many deficits magically disappear altogether. It also gives pension schemes an incentive to hold more equities, to give a higher discount rate and lower liabilities.

2 The “expected returns” approach ignore the cost of the DB guarantee and mis-represents the underlying economics

Using the “expected return on assets” sounds reasonable, after all equities have a higher “expected return” than AA bonds, so doesn’t holding them reduce pension costs?

But the trouble with the “expected returns” approach is that DB are not “expected pensions”, they are guaranteed pensions, and must be paid, even when actual returns are less than “expected”.

Because DB pensions are guaranteed they should be discounted at a guaranteed low-risk discount rate – the AA corporate bond rate.

The “expected return” approach ignores the cost of the DB guarantee. In the real-world this is the cost of buying insurance against underperformance versus the AA return – a “put option” on a stock market index. This gives the right to sell equities at an agreed “strike price” in, say, 30 years, guaranteeing the AA bond return will be achieved.

But because the option strike price would be fixed by AA bond returns, the “expected return on assets” valuation, plus the cost of the equity put option, must always, by definition, equal the value of liabilities calculated at the AA rate.

Calculating DB pension liabilities as a two-stage process – using the “expected return” on assets, plus the cost of an equity put option – does nothing but add complexity, like going from York to Barnsley via Skegness. 

The argument that the “expected return on assets” approach ignores the cost of the DB guarantee is not new. Measuring equity risk by reference to the cost of equity put options was first discussed by Professor Zvi Bodie, of Boston University. in his seminal 1995 paper  “On the Risk of Stocks in the Long Run”   [8]

In 2011 Dr Andrew G. Biggs, Senior Fellow at the American Enterprise Institute, then applied this principle to DB pension guarantees in his lucid paper “An Options Pricing Method for Calculating the Market Price of Public Sector Pension Liabilities” [9]

3 A brief history of pension accounting

Fair market values for pensions – using the AA corporate bond discount rate - reflects the underlying economics of pensions, it isn’t some new accounting fad. It was introduced by the UK Accounting Standards Board in FRS17 [10] way back in 2000, after months of consultation, and later adopted by US and international accounting standards – IAS19. (I advised the ASB on FRS17).

There have been various tweaks to all three standards since 2000, but this has been in the direction of better disclosures, not challenging the use of corporate bond rates.

In the last 20 years global financial accounting in general has moved towards transparency and consistency, putting market values for other costs and liabilities, including share options and leases, in the p&l and balance-sheet.

Market values of financial assets and liabilities for companies, banks, insurers and pension scheme are important in giving transparent free access to the best information for decision making, increasing the overall efficiency of capital allocation in the real economy.

Before FRS17, under accounting standard SSAP24 UK companies used the “expected return” on assets to value both pension liabilities and assets,   which meant the accounting values of liabilities and assets bore no relation to their fair value in the real world. Disclosures were also weak.

FRS17 was certainly a loud wake-up call for finance directors, who now had to recognise and manage pension costs and liabilities in their published accounts, and prompted a step-change in behaviour. Equity and bond analysts could now see what was going on, and asked awkward questions.

Companies started to manage spiralling annual pension costs, first by closing to new members, and then more recently, closing to existing members. They also started to match DB pension assets and liabilities - selling equities, and buying long dated matching bonds, exactly what Boots had pioneered.

Matching assets and liabilities isn’t something driven by an inconvenient accounting convention. It reduces risk in the real world for:

-         Scheme members, reducing the likelihood of assets being less than liabilities, if the employer goes bust

-         The Pension Protection Fund, which provides compensation for members if their employer does go bust

-         shareholders, who have less financial risk “off balance-sheet” in the DB pension scheme

-         the financial system as a whole, since leverage through cross shareholdings” of one company by another, is reduced.

4 What happens if we rip-up AA corporate bond accounting?

The AA corporate bond genie has now been out of the bottle for over 20 years. In practice, I don’t believe much would change in corporate behaviour if we did rip up “AA corporate bond” accounting

Equity and bond analysts, including the credit rating agencies, would certainly waste energy reverse-engineering the published “expected return” pension liabilities to get back to higher market values. And they will still ask awkward questions, and mark-down companies with significant pension risk, regardless of what their published accounts say.

To be consistent many other areas of fair value accounting, as well as pensions, would have to be changed - starting with the value of bonds issued by a company.

Any move away from fair value accounting, and international accounting could be seen as the thin edge of the post-Brexit wedge and undermine international investor confidence, with the pernicious affect of increasing the cost of capital for UK plc.

Ripping up accounting won’t change the flourishing pension buy-out market either, where insurance companies use bond rates to price bulk annuities.

More importantly, although FRS17 changed company behaviour, it was quickly followed by a series of important changes to pension regulation, which also powerfully reinforced this new behaviour. This started with changing the S75 definition of the “debt on the employer” [11] to the buy-out value, massively increasing the cost for any company wanting to wind-up their pension scheme.

The 2004 Pensions Act [12] followed shortly, setting up The Pensions Regulator and the Pension Protection Fund. The Pensions Regulator has  been gradually moving to requiring “fair values” [13], and as a matter of practice, most actuarial valuations use a discount rate as tough, or tougher, than AA corporate bonds.

The PPF  Pension Protection Fund already requires “fair values”, albeit with some (modest) smoothing. [14]

Even if we did rip-up 20 years of pension accounting, the 20 years of  pension regulation would remain in place. Unless we also ripped this up – and there is no suggestion the Committee wants to do this – it is naïve to think that simply changing pension accounting would have much practical impact.

Rather than trying to hide pension risk, we should be making accounting more transparent, so shareholders, bondholders and pension scheme members can see the real extent of pension risk, including Leveraged LDI.

Anyone who wants to fiddle pension accounting should read Winston Churchill’s letters. In 1954 his doctor told him he should not weigh as much as 15 stone, and his wife Clementine tried to put him on a tomato diet, which he resisted.

He solved the problem by testing different weighing machines until he found one that registered half a stone less.

But we won’t solve any of the UK’s pension problems – including Leveraged LDI - by looking for a new weighing machine.