It is a given that any change to pensions accounting will inevitably produce earnings volatility. A Georgia Tech research team led by Dr Charles W. Mulford has recently attempted to answer the question: by how much?

Starting with an analysis of “changes to the balance sheet and its effects on measures of leverage and profitability for the 30 companies in the Dow Jones Industrial Average (DJIA) caused by the initial adoption of US accounting standard SFAS No 158,” The Effects of Enacted and Proposed Pension Accounting Changes on Leverage, Profitability and Earnings Volatility discusses future changes to pensions accounting that could “impact financial statements even further”.

The research also looks at “possible effects on pension expense and income … if full pension costs were recognised in income, instead of flowing through other comprehensive income”.

Implemented in September 2006, SFAS No 158, Employers’ Accounting for Defined Benefit Pension and other Postretirement Plans, marked a significant culture change for businesses reporting under its predecessor, SFAS No 87. It scrapped formerly permitted delayed recognition of retirement and other post-employment benefit costs in financial statements.

What it does not require, however, is the recognition of components such as prior service cost or gains and losses in income. It is here that further reforms to pensions accounting could see even greater earnings volatility - the focus of the second part of the Georgia Tech study.

“It is likely,” the report continues, “that the changes in pensions accounting caused by the implementation of SFAS No 158 are only the beginning of reforms.” Despite the corrections made to balance-sheet accounting, the full fair value of pension plan costs continues to be drip fed into income using a corridor approach.

This state of affairs will strike a chord with those familiar with IAS 19, the Statement of Recognised Income and Expense - SORIE. “A move to fair value accounting,” the report argues, “would call for such changes to flow through income in the year realised.”

In a bid to measure the impact of this change, Georgia Tech analysed 24 companies over a five-year period. They replaced the net periodic benefit cost for each company included in the study with a revised measure. Significantly, the revised measure “now includes actual return on plan assets and the full actuarial gain or loss incurred during the year”.

They selected a five-year period between 2002 and 2006 for analysis. These years, it is argued, capture “years of good and bad economic performance and actual returns on pension plan assets.”

The analysis has two aims: first, to assess the percentage change between the reported and the revised net periodic benefit cost; secondly, to measure the percentage change in income from continuing operations, “assuming the revised pension expense replaced the reported amount”.

The report demonstrates the effect on pension expense where “the full pension costs are included in a given year, including the actual, as opposed to expected, returns on plan assets, rather than the expected return”. The widest variation from the median is reported in 2006, a year which scores the greatest percentage decrease in expense; 2002 sees the greatest percentage increase.

The report’s authors attribute this outcome to strong market returns in 2006 and weak in 2002. They continue that: “In 2002, market returns were weak, leading to an actual loss on plan assets for all 24 companies in the study, when the expected returns were positive for all 24 companies.”

Among the companies studied, DuPont showed a large variation in the adjustment to the pension expense over the five-year period. Measuring its adjusted pension expense as a percentage of its reported equivalent, DuPont’s varies from 2177.0% in 2002 to -374.8% in 2006. The real numbers underlying spell out the point starkly: in 2002, the expected return on plan assets was $1.73bn (€1.12bn); its real-world loss was $1.92bn.

The second feature of the research of interest to those watching the IASB’s IAS 19 project is the analysis of the impact on income from continuing operations of recognising full pension costs. Examining Boeing, in 2002 the researchers found that “adjusted income would have been -177.1% of the reported income”; by 2006, income from continuing operations would run at 167.8%.

To condense the research into one line for those who can’t bear to look: the fairer the value, the truer the picture, the greater the volatility.