Canadian pension accounting rules changed for 2000. Under the old rules, the discount rate was set on a long-term basis and was rarely changed. Under the new rules, the discount rate changes each year as long bond yields change. As a result, pension expense under the new rules will be significantly more volatile.

This article presents a case study describing how Company XYZ implemented the new accounting rules. Company XYZ is primarily in a regulated business, and wanted to implement the new rules so as to minimize year-over-year fluctuation in pension expense. Their plan is a traditional final-pay design. Their asset allocation, set by the Investment Committee of the Board, was 75% fixed income and 25% equity. The client was the Finance & Administration Committee of the Board. The F&A committee is responsible for the Company's financial statements, but the Investment Committee has responsibility for asset allocation.

Results
We explored with the F&A Committee the following implementation questions for the new accounting rules:

* Should the experience corridor be used? This allows for accumulated experience gains and losses that are less than 10% of the greater of the assets and liabilities to be ignored when determining pension expense.

* How should the salary increase and return on asset assumptions be set? For the long term, or pegged to the discount rate and changed every year?

* What method should be used to determine market-related value of assets?

* What is the impact on pension expense volatility of increasing the equity allocation or lengthening the bond duration?We helped the company to answer these questions by modelling the financial results of their pension plan over the next 10 years on a stochastic basis. Specifically, we looked at the year-over-year volatility in pension expense under alternative expense policies.

As expected, the corridor significantly reduced the volatility of pension expense and was used at implementation.

The impact of fixing the salary increase assumption for the long term, as opposed to having it float with the discount rate, was minimal on pension expense. Their auditors had a strong preference for infrequent changes in the salary increase (and return on asset) assumptions. As a result, the company chose to keep their salary increase assumption fixed for the long term (reviewed annually, but changed only infrequently).

The asset smoothing technique that the company had used under the old accounting rules did not produce a smooth pattern of pension expense under the new rules. Because the liabilities are directly affected by long bond yields and the fund's asset allocation was heavily weighted to bonds, using market value generated a significantly smoother pattern of expense. Market value was adopted on implementation of the new accounting rules. The company did understand, however, that this decision should be reviewed following any material change in their asset allocation.

Increasing the equity weighting of their asset allocation increased the volatility of pension expense (but also reduced its level). Since the low equity weighting produced relatively less volatile expense patterns and asset allocation was not the responsibility of the F&A committee, no change was recommended in the debt/equity mix of the allocation.

Increasing the duration of the bonds (from roughly six years to almost 10 years) significantly reduced the volatility of pension expense. In addition, it reduced the expected level of pension expense somewhat. We recommended that the F&A committee discuss the bond duration with the Investment Committee, with a view to lengthening the duration.

Conclusions
Pension expense is making an increasingly important impact on the sponsor's financial statements. The only way to assess the impact of the volatility of pension expense is by examining the results of a stochastic asset/liability projection of the finances of the pension plan. Taking this approach also provided the plan sponsor with a deeper understanding of several of the levers that may be used to control the level and volatility of pension expense in the future. A similar approach could also be used to illustrate the effect of other available levers, such as contribution policy.

Steve Bonnar is a consultant and principal at Towers Perrin in Toronto.