[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries
.

Public sector pensions

Public sector pensions are the ‘gold standard’ of pensions, and those who are in these schemes guard them jealously. The equivalent defined benefit (DB) pension schemes in the private sector have been closing to new entrants, and indeed some (although still a small minority) have closed existing DB schemes completely, forcing employees into a defined contribution (DC) scheme for future accrual. Private sector schemes have been closing because their sponsoring firms believe they are too expensive. But if they are too expensive for the private sector, why are they not too expensive for the public sector?
The answer to this question is that public sector employers do not know how expensive their schemes are. The reason for this is that they are not charged enough for them, nor are they responsible for paying the pensions.
The reader may find this puzzling how can a public sector employer (say an NHS trust) not know how much its pensions really cost? To keep the argument simple, I will illustrate using only unfunded public sector schemes as examples, not funded ones.

Unusual structure
To understand how public sector pensions work (and they are very unlike private sector pensions), we have to understand how they are organised within the various parts of the public sector. The employers (the civil service, NHS trusts, local education authorities, the military, police authorities) do not pay pensions to retired staff. These are paid by the Treasury, via pensions agencies (eg NHS pensions are paid by NHS Pensions, which is part of the NHS Business Services Authority). These agencies act as the administrator and paying agent, but the money comes directly from the Treasury, not the employer. In its turn, the Treasury charges the employer for the pension, but not for the same amount as it pays out. The Treasury charges the employer an amount it believes is necessary to cover the current value of the extra cost of pensions accrued in the current year’s employment (service cost). This is a payment which is supposed to cover payments of pensions many years in the future. The Treasury then pays out whatever the contractual obligation of the employer is to its current pensioners ie it takes all the financial risk. At the moment, the aggregate amount charged to public sector employers under current methodology, and the aggregate amount paid out to public sector pensioners, is not very different. So the Treasury does not see a large difference between current money in on public sector pensions, and current money out.

Methodology
The Treasury charges the employers a proportion of salary for example, for NHS staff, this is a total of 20% of pay, of which on average employees pay 6%, and the NHS trusts and other NHS employers pay 14%. So the questions are:
– where does the 20% number come from, and
– is it enough to cover the future pensions liabilities of the Treasury?
The answer to the first question is that it is arrived at by an actuarial calculation of how much has to be put aside today to ensure that if it were invested in index-linked gilts, it would be enough to cover all future pensions, including increases in future longevity and earnings (to which pensions are linked before retirement). This calculation cannot be exact (since longevity and earnings change), but an actuary can make a good stab at forecasting future changes, and incorporate these into the current cost calculation. Indeed all the main public sector schemes publish annual accounts (called resource accounts) which include an actuarial report, and many use a model called SCAPE (‘superannuation contributions adjusted for past experience’), developed to cope with the unusual situation of valuing a pension promise with no fund. The theory is unimpeachable the technique asks the question ‘what value of (notional) index-linked gilts does this fund need at date X to fully cover all the future liabilities in the pension scheme valued at date X?’. This technique will also allow the calculation of the annual service cost using conventional actuarial methods.
Clearly there will be some other important variables that will affect the calculation through time (mortality estimates and trends, and future earnings assumptions for active members). While I have some reservations about the current assumptions for these variables in the main public sector schemes, the impact of changes to these assumptions are small compared to the discount rate assumption.

Discount rates
The calculation relies most crucially on the real (ie inflation-adjusted) interest rate applied to the index-linked gilts which is, under SCAPE, the same thing in effect as the real discount rate applied to the future liabilities.
The government had, up to 31 March 2005, insisted on using an artificial investment return of 3.5% pa over inflation (the ‘real’ interest rate). Since 1 April 2005, it has used a slightly lower, but still artificial, real rate of 2.8% pa. But actual real interest rates (ie those obtainable in the market) have been much lower than this for many years. In March 2005 the market real interest rate was 1.6% pa, and in March 2006, it was 1.1% pa. The government knows this, because it is currently exploiting these (historically low) interest rates to issue long-term debt at these fixed real rates.
This may sound academic, but it is not. If we calculate from first (ie SCAPE) principles the annual cost of, say, a typical NHS pension at 3.5% pa real return, we get remarkably close to the Treasury’s annual charge to the employer of 20%. If we recalculate the correct charge at the prevailing interest rate at March 2006 (1.1% pa), the amount that employers should be paying to the Treasury is about 38% of pay! (See figure 1.)
So if, say, the NHS trusts were forced to run their own pension schemes, and set aside enough money to fully fund them (or notionally fully fund them) by investing in risk-free government investments at prevailing market interest rates, they would face an increase in their own contributions from 14% of pay to 32% of pay (assuming they did not charge the employees any more), or an increase in the wage bill of 18% overnight!


Choosing the discount rate
The usual argument against using risk-free interest rates for this calculation is that pension funds can invest in higher-yielding (but risky) investments like company shares, corporate bonds, and property. Hence the argument is frequently made that liabilities should be discounted at a rate that is higher than the risk-free rate.
The problem for all public sector schemes (apart from the local government pension scheme) is that these schemes are unfunded that is, the Treasury has not put the money aside. Instead, it has spent it. So there is no money to invest, and no higher returns available. Indeed, the question of the appropriate discount rate is in my opinion scarcely a matter for debate in this unfunded environment, since the Treasury can be thought of as deferring the issue of gilts, which it would have had to make had it not had the pensions’ contributions from the public sector employers. Since gilts would have been issued at the market rate, such a deferral can be seen as an investment in gilts (index-linked or otherwise) also at the market rate.
Interestingly, the artificial real rate chosen by all the public sector schemes is done so on the advice of a body called the Financial Reporting Advisory Council (FRAB), which in turn takes advice from the Treasury and the Government Actuary’s Department (GAD). Of the two rates FRAB has recommended (3.5% pa and 2.8% pa), the former is based on ‘a review of long-term historical patterns of real rates of return on gilts’ (source: FRAB Seventh Report; section 2.10), the latter based on its interpretation of the FRS17/IAS19 AA corporate bond rate. The move to the lower rate will increase the reported outstanding unfunded public sector pension liability from the latest (March 2005) value of £530bn to about £640bn for March 2006. Note that this £640bn estimate assumes the other main assumptions remain the same. However, it is possible that the future real earnings assumption will be upped from 1.5% to 2% pa this seems to be presaged in a recent document from GAD (‘Unfunded Public Service Pension Schemes 2006 Cashflow Projections Methodology, assumptions and data’, 26 January 2007.
This estimated value has not yet been announced at the time of writing (April 2007), and this more-than-12-month delay in publication is unusual, and may indicate that the government sees even this number as politically embarrassing. However, using the methodology above, the March 2006 liability figure using March 2006 market interest rates is £1,025bn, or about 83% of 2005/06 GDP.
The further implication of the modest reduction in the discount rate has not yet fed through to contribution rates. For consistency, the Treasury should have charged public sector employers a higher contribution rate from 1 April 2005 and using figure 1 as a guide, the charge should have been around 24% rather than 20%. Note this is just for consistency with the government’s own accounting not the correct amount, as we have seen above. To reiterate, the actual, market-based figure, would be an average of around 38% of salary.

An economic rate for pensions
The government is facing an increasingly uphill battle to preserve this indefensible discount rate. My guess is that the weight of opinion, taking its lead from changes now in train in the private sector, will force the government to move towards market-based pricing of its pensions. When this happens, it is likely that it will charge public sector employers the economic rate for pensions (which it thinks it is doing at the moment, but is not). Once public sector employers face charges this high, the same pressures that applied to private sector schemes will apply to them. Many will then say that pensions are ‘too expensive’ and close their final salary schemes.

07_05_04.pdf