A Brief Review of the DWP’s Impact Assessment of the proposed Occupational Pension Schemes Funding and Investment Strategy and Amendment Regulations

Iain Clacher & Con Keating

The Impact Assessment is fifty pages long; much of which is concerned only with minutiae. However, the headline conclusion is surprising:

And contains some limited further information:

We find this conclusion and these values extremely surprising. For example, the cost of moving the discount rate from gilts + 150 basis points[1] to gilts +50 basis points when applied to the existing stock of DB claims is £172 billion.

We have modelled the existing liabilities of schemes on the basis that they all move within ten years from March 2023, and take a variety of paths to achieve this transition to a low dependency asset allocation. We would also note that there is a very significant difference in the estimates of scheme assets at that date. The ONS report £1.244 trillion, TPR £1.425 trillion and the PPF variously £1.439.8 and £1.404.4 trillion.

The increase in the cost of liabilities to schemes in aggregate varies from £84 billion to £245 billion.

We would note that prior to March 2022 there was little difference between the ONS estimates of total assets and those of TPR and the PPF. As we have highlighted elsewhere, it would appear that the latter’s estimates fail to capture the effects of the LDI crisis, and the PPF is explicit in the fact that its methodology does not capture leverage, repo, and portfolio rebalancing all of which were material over the course of 2022.

In the impact assessment there is no estimate of the savings to the pensions of scheme members arising from the higher levels of scheme funding, for schemes which experience insolvency. In any cost-benefit analysis, this should be central.

The assessment states:

“Other key non-monetised costs by ‘main affected groups’

There may be a cost to pension savers. Members of open cost-sharing pension schemes may see some of the costs associated with the regulations passed on to members, either directly (higher contributions) or indirectly (such as wage impacts). However, as only 10% of schemes are still open to new members; cost-sharing schemes are a small minority, therefore we do not anticipate any material impact on members in aggregate because of this. There may be further costs to TPR in relation to monitoring compliance.”

We have modelled this on the basis that all schemes are closed to new members and future accrual. We show below as Table 1 the average ‘savings’ for a range of insolvency rates, and a loss experience of 13% of the projected pensions, together with the loss experienced by scheme members.

Table 1: Loss savings under low dependency funding for a range of annual insolvency rates

Note that these are present values of all future losses in insolvency.

As such, they are also sensitive to the discount rate applied. For example, in the case with a 1% insolvency rate,  the savings fall from £209 billion to £168 billion with a 100 basis point rise in the discount rate, and increase to £261 billion with 100 basis point fall in the discount rate.

It is also worth considering the rate of decline of risk exposure, which is shown in Figure 1 below. The loss exposure has halved by year 7.

Figure 1: Loss exposure (4.1% discount rate, 1% pa insolvency)

The assessment summary also contains the following table.

We presume that the number ten shown in the heading of transition costs refers to the cost of transitioning over ten years. However, even assuming these are annual figure over this ten-year period, the values for transition costs are implausibly low. For example, USS has estimated its transition costs for a ten-year period at £6-£8 billion pounds.

We have been unable to estimate the costs for sub-groupings, such as the 1200 referred to, but again it really does not matter as the £7.1 billion cost referred to in the text is implausibly low. We would note that the shift quoted earlier from gilts +150 to gilts +50 would imply that schemes would need to have been previously funded at 133.4% in order to not be in deficit after that shift, and very few were.

We cannot reconcile these annual figures or even the ‘Total Costs’ as present values using a common discount rate.

The assessment summary also shows a benefit table:

The similarity, but not equality, with the costs table is striking but also unintuitive.

Previously, schemes would have to have had large surpluses to not now be in deficit. It is feasible that 1400 schemes could be in surplus after the transition if the PPF asset estimate of £1.4 trillion is used, but it is not if the ONS estimate of £1.244 trillion is applied.

It is also noteworthy that in several places the assessment prays in aid questionable PPF statistics. For example, in clause 15 it contains:

“This is demonstrated by the PPF 7800 Index with funding ratios increasing from 111% (March-22) to 133% (March-23) again on a s179 basis.”

If the ONS asset estimate is used rather than the PPF estimate, the funding level reported in March 2023 would have been 111.5%.

The Intended Purpose of the Regulations

Clause 22 outlines the intended purpose of the Regulations, and contains, inter alia, the following with respect to the sponsor employer:

• Ensuring funding and investment risks are supportable. Regulations would link the maximum level of funding and investment risk schemes can take – primarily – to the employer covenant, as well as the maturity of the scheme. This is particularly important given the size of DB assets and the time until pensions are paid out getting closer for the majority of schemes. Additionally, this aims to avoid an investment risk spiral, whereby mature schemes invested in growth assets may face market volatility. If there are large asset losses due to market falls, the scheme may not have time for asset prices to recover to pay full pension benefits, and thus invest in further riskier investments.
• Schemes will have less reliance on employers. The regulations require schemes to reach a point of low dependency on their employer by the time they are significantly mature. This will limit the need, or expectations, that additional employer contributions will be needed from that point onwards.

It is evident from these objectives that the cost to the sponsor employer and in aggregate UK plc are really not considered. , The impact that large deficit repair contributions can have on the ability of the sponsor to make investments, develop their business, and contribute to economic productivity and growth more widely, is almost non-existent in the analysis. This is directly in conflict with the government’s investment and growth agenda, with the sole recognition of this issue being acknowledged in clause 39:

“Sponsoring employers’ sustainable growth must be considered when assessing when they can reasonably afford to recover a deficit.”

These regulations will have negative impacts on the population at large including DB scheme members. Indeed, it seems likely that the benefits accruing to scheme members from these Regulations will be less than the harm done to many, and possibly most of them by the wider economic consequences of lacklustre investment, productivity, and economic growth.

While it is extremely unfortunate that the final version of TPR’s Funding Code was not published simultaneously with the Regulations, the emphasis on sponsor covenant and particularly the short-term nature of earlier versions of that Code further reinforce the conflict with government objectives. The only way that this would not conflict with the government’s the case were if all deficit repair contributions are invested in growth assets in the real economy..

Clauses 32-34 consider the options available to DWP, the first of these, do nothing, is clearly superior to the others once the true magnitude of DRCs under the self-sufficiency approach is recognised. We would also point out that measuring liabilities on the basis of a low self-sufficient discount is a prime example of reckless prudence. Given the emphasis that TPR places on low-risk investment and asset allocation strategies and their over-emphasis on buyout, it can be expected that asset allocations will be far more conservative as a result of this policy. In the absence of these matching strategies, for the majority of schemes, the result over time will be substantial trapped surpluses as a result of over-funding, which are far larger than the benefits accruing to scheme members.

Clause 39 does contain:

“Open schemes can take new members into account when calculating their maturity, which will extend the time before they are expected to begin to de-risk and for those schemes that are open to new members and who remain truly stable, then they will not be expected to mature over time in any event.”

The absence of future accrual from this concession is a concern which is not fully resolved by the subsequent paragraph in this clause. Fully open schemes are 9% of the total while a further 36% are open to future service awards.

“To make clear that the determination of significant maturity can include an assumption for future accrual and new entrants, no specific limit is placed in the regulations in order to afford more flexibility to open schemes. However, the regulations make clear the assumption must be based on the covenant of the employer.”

The assessment contains the following summary:

The figures shown in this table are simply not credible. However, this is not the primary concern; that lies in the elements not monetised, notably the benefits and costs to members. These benefits are precisely what was used to justify the introduction of these Regulations. There is also no evaluation of the costs to the wider population and economy.

The analysis in the assessment shows in Clause 88, the overall balance sheet of schemes, with the following introduction:

“Table 2 shows the starting position of the DB universe under the counterfactual and central positions (along with the sensitivity). As can be seen, the asset positions are the same reflecting no adjustment is made to asset values at the start. There is a small difference in liabilities under the Central approach reflecting scheme levelling up are broadly equal to schemes levelling down – though it is important to note the sensitivity with around +/- £12bn on higher/lower scenarios.”

And Table 2 is reproduced below:

In other words, the analysis does not consider the effects of the increases in interest rates and the losses of asset values sustained from this, and in particular LDI-related losses. Figure 2 below shows the asset estimates of TPR, ONS, our own estimate (T) and the two PPF estimates.

Figure 2: Overall Scheme Assets as estimated by TPR, PPF, ONS and our own modelling for the period December 2021 until March 2023

In Figure 3, for the ONS liability estimate, we show our estimate of the present value of full scheme liabilities using a common discount rate to that of the PPF. The ONS does not yet produce liability estimates, but we understand that these are likely to become available from the ONS shortly.

Figure 3: PPF liability estimates and our estimate of full scheme liabilities using a discount rate common to the PPF

Our Table 2 below shows the assets and liabilities of schemes, showing in the first column, the PPF figures for s179 liabilities, in the second column the ONS asset estimate and in the third column the ONS estimate of assets together with a discount rate common to PPF figures. The highly publicised PPF 7800 index funding ratio of 133.2% falls to 115.1% when the ONS estimate is used, and finally to 99.9% when a common discount rate estimate is used for liabilities together with the ONS asset value.

Table 2: Assets and Liabilities of Schemes as at March 2023

As can be seen from Figure 3 above, there was a substantial downward revision of liabilities over the course of 2022, which were unrelated to changing actuarial assumptions. These revisions dwarf the variation of liability values reported in the Impact Assessment’s Table 2 above.

Other Concerns

There are numerous other elements of the assessment which concern us. For example, Clause 81:

The regulations set the framework for the DB scheme funding code to provide parameters to help guide schemes which may change the calculation of important assumptions, including:

• Discount Rates – As liabilities are a flow of pension payments due in the future, these need to be discounted to the present day by a discount factor reflecting the assets held to pay the benefit payments. A higher discount rate may show an improved funding position, as it leads to lower liabilities, but may not necessarily be representative of a well-funded scheme.

• Maturity Point – It is expected that as schemes mature, less risk will be taken and more conservative assumptions (discount rates) used. The regulations will determine when “significant maturity” is reached.

• Risk – As schemes mature, there is an expectation investment will move from growth assets (such as equities) into safer assets (such as government bonds). This will lower the discount rate (hence higher liabilities), as assets will be expected to provide a lower future return.

• Long Term Objective (LTO) – Helping schemes plan what the long-term objective of the scheme should be, for example such as reaching buy-out (where an insurer buys the scheme and the employer is no longer responsible), or alternatively running the scheme on in a low-risk manner.

In the absence of TPR’s Code, these parameters are a matter of speculation.

There are also some strange and incorrect assertions, such as in Clause 97:

“TPR modelling suggests that the improved funding from higher liabilities led to a reduction in counterfactual DRCs by broadly 50% when compared to the position as at 31 March 2021.”

If funding has improved, it has done so from lower present values. The effects of the rate hikes and LDI crisis were, because of the mathematical denominator effect, to lower the funding ratios of schemes in deficit, and such schemes would have DRC schedules in place, and for schemes in surplus, it would further their surplus, and such schemes of course did not have DRC payment schedules in place as they were already in surplus.

Similarly, for Clause 103:

“The Regulations make clear that the trustees of most DB schemes are potentially taking less risk than will be required by legislation (for example around 70-75% of schemes satisfy TPR’s Fast Track parameters in relation to investment risk and are broadly the same in relation to technical provisions). Our analysis estimates that the Regulations could provide a greater incentive for around 1380 schemes to invest more productively, which may help unlock up to £5bn of further investment in private equity and venture capital.”

If this is the case, there would appear to be very little justification for these Regulations or TPR’s forthcoming code. It is also worth noting that the unlocking of £5bn for PE and VC seems to be pitched as some great positive, but in the context of PE and VC, this is a drop in the ocean.

Summary

Simply put, we do not believe that this impact assessment is fit for purpose. Moreover,  these Regulations will bring substantial further costs onto employer sponsors, costs which far exceed any benefits they may bring to pensioners and scheme members, and at a huge cost to the real economy, productivity, investment, wages, and economic growth.

[1] The assessment assumption of current discount rates for fast track modelling is actually higher at gilts +200 basis points.