Accounting for Pensions Survey 2000 
Detailed analysis of Reports and Accounts


Actuarial assumptions



SSAP 24 does not specify what assumptions should be disclosed referring only to "a brief description of the main actuarial assumptions". It is widely accepted that full disclosure should include the rates of:

  • investment return

  • salary growth

  • pension increases

  • dividend growth (where this is used to value assets)

The assumptions can either be disclosed as absolute amounts, for example:

  • investment return 7½%

  • salary growth 5%

  • pension increases 3%

  • dividend growth 3½%

or else as differences, for example:

  • salary growth 2½% below investment return

  • pension increases 4½% below investment return

  • dividend growth 4% below investment return

or a combination of the two.

For the 83 companies who were analysed in detail the assumptions were disclosed as follows:

  • absolute amounts                   56

  • differences (or combination)     27

We generally find disclosures where assumptions are expressed in absolute terms easier to follow than those where assumptions are expressed as differences or, even worse, as a combination of the two. We think that companies should reveal the absolute amount of each assumption.
ACT effect Despite the fact that the ACT change was as long ago as July 1997, 18 companies based their pension cost on actuarial valuations which pre-dated the ACT change and took no account of it. This was an important factor in assessing the assumptions used, and we have therefore separately analysed the pre-ACT and post-ACT assumptions.
Investment return The level of investment return assumed depends to some extent on the actuarial method employed. Where a traditional method is used, under which assets are valued by discounting the income they are expected to produce, the investment return is the long-term expected return.

Where a market value basis is used for the assets, the investment return should be determined from market conditions at the valuation date and will very often differ from the assumed long-term return used in discounted income valuations. The returns below therefore exclude the eleven companies using market value assumptions.
Split
investment
returns		 Sometimes the long-term investment return is divided between one return before retirement and another after retirement, reflecting a prudent view that as the liability for pensions in payment increases it will be matched by investment in bonds instead of equities. In such cases we have used the pre-retirement investment return. This was also used to determine the gap between investment return and salary growth, whereas the post-retirement return was used in assessing the gap between investment return and pension increases.

The range of investment returns assumed is shown in the charts below.

Chart: Investment returns for companies using pre-ACT assumptions

Chart: Investment returns for companies using post-ACT assumptions and a non market basis
Timing effect Overall there has clearly been a downward shift in investment returns. Even within the post-ACT set, the assumed investment return has been reducing over time. The average investment return for valuations after the end of 1998 was 7.3%, nearly ½% less than the average of 7.7% for earlier valuation dates.
Investment return over salary growth The rate of salary growth assumed can, and does, have a significant effect on the eventual pension cost. The smaller the "gap" between the assumed rates of investment return and salary growth, the more conservative the assumptions.

The difference between the investment return and assumed salary growth could be analysed for 83 companies. The difference varied from 1% to 4%. There was, in this case, no significant difference in the spread of assumptions between the pre-ACT and post-ACT assumptions.

Chart: Investment returns over salary growth

It should be noted that some companies will have allowed for promotional salary increases in addition to the general increases to which their disclosures relate. Few companies disclose details of promotional salary scales and this makes comparison more difficult.
Investment return over pension increase 73 companies disclosed sufficient information to enable the difference between the assumed rates of investment return and pension increase to be determined. Here too there was no noticeable variation between the pre-ACT and post-ACT assumption ranges.

Chart: Investment returns over pension increases

One might have expected to see a downward shift in the gap between assumed rates of investment return and pension increase reflecting lower real investment returns. However, there has been no noticeable trend.

It is important to note that the assumption made for pension increases will reflect the increases actually awarded under the scheme. An assumption that the investment return will exceed pension growth by 5% pa may be judged conservative where a scheme pays no increases, but optimistic if the scheme guarantees increases in line with inflation.

The Pensions Act 1995 requires companies to index-link pensions built up for service after 6th April 1997 (up to a ceiling of 5% pa). This is known as Limited Price Indexation (LPI), and could well be different from the rate of increase provided for all pre-April 1997 pensions.
Assumptions used to value assets The value of the assets is highly sensitive to the difference between the investment return and the dividend growth rate. For example, a reduction in this "gap" from 3½% to 3% would increase the equity part of the asset value by about 15%. This will increase a surplus or decrease a deficit, and the change is spread forward over the average future working life of employees resulting in a reduction in pension cost. For a mature scheme a ½% change in assumed dividend growth could, for example, halve the disclosed pension cost.

44 companies using the discounted income method gave sufficient information for us to determine the difference between the assumed rates of investment return and dividend growth.

The dividend growth assumptions showed a very strong difference between the pre-ACT and post-ACT assumptions, as the two following charts illustrate.

Chart: Investment returns over dividend growth for companies using pre-ACT assumptions

Chart: Investment returns over dividend growth for companies using post-ACT assumptions

Last year's survey, where more pre-ACT companies were included, showed a majority of companies using a "gap" of a little under 4.5%. The post-ACT assumptions this year show the gap is generally reduced to between 3% and 4%. On average the pre/post ACT difference is close to ½%.

The direct impact of ACT is to reduce the dividend income being valued by a little under 20%. As explained above, a reduction of ½% in investment returns over dividend growth will provide a compensating increase of around 15% in the value of equities. Overall, therefore, we believe that actuaries have changed their assumptions, and this to reduces the immediate impact of the ACT change.
British Aerospace British Aerospace valued the assets of the main pension schemes allowing for dividend growth at 5% (or more) below investment returns. This is comparatively a very conservative assumption. The 5% gap is about 1¼% higher than the average. This could have reduced the overall funding level by nearly 20%. At the same time they disclose a funding level of only 95%. Thus shareholders might be concerned about an apparent underfunding when, on a basis used by many other companies, the pension schemes are well over 100% funded. There is also a knock-on effect on pension cost; a 20% asset value increase on the principal schemes could have decreased the pension cost (and increased profit) by around £100 million, or over 10% of pre-tax profits. There may be other factors which show that the pension costing basis overall is not as conservative as we suggest, but they are not disclosed in the company report.
Market value methods Eleven companies used a market value method. As explained in section 3, earlier, the inflation rate assumed appears to have been derived from returns on gilts. The rate of assumed salary increases, relative to the investment return, showed a similar range as for the traditional actuarial valuation methods.

The investment return can in theory be linked to the return on index-linked gilts plus an "equity risk premium", ie the extra return expected from equities for their risk relative to government stock. Analysed in this way, the equity risk premium ranged from -½% to 2.4% with no discernible pattern. It seems that the actuarial world has yet to arrive at a consensus on market value assumptions!
Imperial Tobacco Imperial Tobacco used very conservative market value assumptions. The investment return was taken to be just 2.4% higher than inflation and only 0.4% higher than salary increases. This was at a time when a return of 2.9% pa above inflation could be guaranteed using index-linked gilts!

Despite the conservative basis, the UK pension cost was nil. Were the assumptions deliberately prudent to avoid a pension credit in the accounts?
Back to top