[Skip to content]

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

Pensions: Accounting for risk

We know that defined benefit pension schemes are risky: disappointing investment returns and rapidly improving longevity are among the factors that have helped produce substantial deficits in some schemes. One outcome has been firms having to increase their contribution to schemes, disrupting planned cash flow, interfering with mergers and acquisitions and, in some cases, increasing fears of insolvency.

Our research examined how firms disclosed in their accounts the risks to which they are exposed from their pension obligations. The International Accounting Standards Board is also reviewing its rules in this area. Our study covered the 88 firms in the FTSE-100 as at 31 December 2009 that have a defined benefit pension scheme. We looked at what they disclosed about pension risks in their 2009 accounts.

From the results, we were able to suggest an alternative form of disclosure, where firms concentrate on the risks arising to it from their pension obligations. Actuaries reporting to scheme trustees may also wish to review how they report on risks.

Our benchmark was the Reporting Statement of the Accounting Standards Board (ASB), issued in 2007, which suggested that firms disclose how sensitive the value of pension liabilities is to alternative assumptions on price inflation, salary growth, expectation of life and the discount rate. An example was that a firm could disclose that a 0.5% increase/ decrease in the discount rate would lead to a (say) 9.5% decrease/increase in the liabilities. This links in with International Accounting Standard 1, which refers to firms explaining where their estimates of assets and liabilities are uncertain. This does not apply to items that were fair valued on the basis of recent trades, since they would not (at least normally) be regarded as uncertain estimates.

We can identify noticeable differences in the assumptions that firms make, which is one indicator of uncertainty. Restricting ourselves to firms with a balance sheet date of 31 December, and excluding firms if they had no scheme in the UK, the price inflation assumption varied from 3.1% pa to 3.9% pa, and the discount rate from 5.5% pa to 6.0% pa. While the ranges are partly attributable to differences in firms’ circumstances, such as the duration of liabilities, they also reflect the difficulty in deriving assumptions for a long-term liability.

We examined the extent to which firms disclosed the sensitivity of the liabilities to the four risk factors listed by the ASB. The result is shown in Figure 1, which shows that, on average, only 1.6 out of the four disclosures are being given. Disclosures were more likely if the pension liabilities were large (greater than the median in Figure 1) or well funded. The ASB Reporting Statement is, however, only a suggestion of best practice rather than a rule.

However, the design of sensitivity tests does raise a number of questions; in particular, what magnitude of change in the risk factor should be used? If it is meant to be an alternative that was reasonably possible at the balance sheet date, then a 0.5% change in the discount rate looks unduly large. It is therefore not surprising that, of the 52 firms disclosing the discount rate sensitivity, 34 used a change of less than the ASB’s 0.5% figure.

Another area where there was variety was that some firms reported the effect of an ‘increase/decrease’ as referred to above. However, others showed the effect of an increase and a decrease separately, while some showed either the effect of an increase only or a decrease only. Although firms did have different approaches in detail, we estimated what the sensitivities would be on the basis illustrated by ASB; for instance, that the effect of a 0.5% increase/ decrease in financial variables (and a one-year increase in the expectation of life). The outcome is shown in Table 1. The differences between firms suggest there is merit in better quality disclosures of pension risks.

However, our main concern is that the disclosures mostly refer to what we call ‘point in time estimation risk’, ie. the uncertainty surrounding estimates at the balance sheet date. ‘Enterprise risk’ is also important: what are the risks to the firm from the pension scheme and how are they managed? In this context, the assets of the pension scheme are important as well as the liabilities; and the magnitude of change in the risk factor should reflect what is reasonably possible over perhaps a five-year period.

From this perspective, the main risks are interest rates, share prices, price inflation, salary growth and longevity. One approach to this question could be to disclose Value at Risk (VaR). For example, it is 95% certain that there will not be a deficit of more than £X over a stated period. We found no pension VaR disclosures in firms’ accounts. Indeed, recalling the criticism of banks’ VaR calculations, this is perhaps not surprising.

However, sensitivities can be effective in disclosing enterprise risks. For example, if there were a 0.5% change in interest rates, by which we mean a parallel shift in the yield curve for all bonds, what would be the scheme liabilities (with discount rates changing by 0.5%) and to what extent would the assets change? We could then identify the extent to which the investment strategy was sheltering the firm from interest rate risks. At present, this is difficult to ascertain from the accounts, since there is usually no information about the duration of the bonds, or the make-up of liability-driven investment portfolios. Similarly, we would like to see the effect, on not only liabilities but also assets, of changes in price inflation (since there may be RPI-linked securities) and changes in longevity (there may be insurance policies or longevity derivatives).

How important are pension risks? Using the imperfect information in the accounts, we estimated that the median impact of a 0.5% reduction in interest rates on assets and liabilities and hence the book value of firms’ equity was 1.9%. For a 20% fall in the value of equities, the median effect was 2.5% of the book value of equity. In some firms, the pension scheme is not an important risk factor, and lengthy compulsory disclosures are inappropriate.

We suggest, at least for large schemes, it would be useful if firms showed the effect of an increase and decrease of:

>> A 0.5% pa change in price inflation with salary growth, real salary growth and interest rates
>> A 20% movement in share prices
>> A one-year change in the expectation of life on each of the following: — Pension assets — Pension liabilities — Pension service costs in the income statement.

We believe this approach will enable readers of accounts to understand the main risks to firms arising from the pension scheme and the extent to which they are being managed. There would need to be an accompanying narrative disclosure, but this can be more concise if we have sensitivity disclosures on the lines suggested above.

The report is available at www.nottingham.ac.uk/business/cris/papers/Pension_risk_disclosures.pdf

We are grateful to the Institute of Chartered Accountants of Scotland for funding this research.

___________________________________________________________

Chris O’Brien is director of the Centre for Risk and Insurance Studies at Nottingham University Business School, Margaret Woods is reader in accounting at Aston Business School and Mark Billings is lecturer in accounting and risk at Nottingham University Business School