"Efficient Gain and Loss Amortization and Optimal Funding in Pension Plans," M. Iqbal Owadally and Steven Haberman, January 2004

North American Actuarial Journal, Apr 2004 by Gold, Jeremy, Cowling, Charles, Exley, Jon, Hudson, Nick, Et al

We also agree with the suggestion in section 1.4 that "Another motivation for pre-funding pension benefits is to spread the required future pension contributions and hence reduce strain on the sponsor's cash flows." Corporations seek to avoid large pension flows at the same time as tight cash flows elsewhere in the business. Corporations might also worry about the impact of pension risk on their cost of raising capital to finance their core business. Variability of pension flows themselves is a small part of the picture. Variable funding contributions might even be a good thing if they naturally hedged some other aspect of corporate cash flow. Wise corporate risk managers are not concerned with the variability of pension flows in themselves, but instead with the effect of a pension plan on an already variable corporate cash flow.

As with the funding level, a covariance is required, but the authors argue instead that "minimizing the variability of the contribution ... is also a reasonable objective." Their assertion would be valid only if plan cash flow were independent of other corporate cash flows. The authors escape building a corporate model only by a heroic stochastic independence assumption regarding the cash flows of the pension plan and the sponsor.

It is surprisingly straightforward to analyze pensions even when a plan is correlated with the sponsor's fortunes. We characterize the emerging literature as "business-centric" rather than "plancentric." Only arbitrage arguments and clear thought are required in order to understand the key conclusions. No calculations are needed. Sharpe (1976) and Treynor (1977) set out the main findings. Non-zero correlations have a profound impact-implying that the same risk controls routinely applied to corporate balance sheets are also appropriate for pension plans.

Some difficult questions relating to discretionary benefits, default risk, and tax require more detailed modeling. Clear thinking and model complexity need not exclude each other. For example on the question of amortization, Chapman et al. (2001, section 6.9) deduce the following:

"Shortening the [amortization] period [from 10 years] to 5 years increases the percentage of scenarios where the company goes into liquidation . . . Note that both shareholders and employees are winners as a result of this change. . . . The losers are the Government, which receives less tax, . . . and holders of debt, who suffer a higher rate of default."

Their approach is more useful for decisions than the corresponding findings of Dr. Owadally and Professor Haberman (see section 2). "Dufresne (1988) concludes that . . . spreading gains and losses over a period between 1 and 10 years is efficient. . . . Owadally and Haberman (1999) conclude that the . . . practice of amortizing gains and losses over 5 years is also efficient."

The results of Chapman et al. (2001) are consistent with those of Sharpe (1976) and Treynor (1977), but conflict with the claims of Dr. Owadally and Professor Haberman. The current authors' refusal to cite the business-centric literature is all the more remarkable because a written discussion of Chapman et al. was published in the British Actuarial Journal (IIaberman 2001).