MEASURING TERMINABLE POSTRETIREMENT OBLIGATIONS

North American Actuarial Journal, Jan 2005 by Petertil, Jeffrey P

ABSTRACT

New approaches are needed to value benefit plans subject to unilateral changes or termination. The paper focuses on postretirement health benefits, but the thesis may be relevant to any flow not guaranteed by law or accumulating funds.

Retiree health benefits have usually been extended to participating active employees only in concert with a reserved right by the plan sponsor to control the design and, by implication, the cash flow. Over the course of the last fifteen years, this reserved extension of benefits has almost invariably led to reductions in benefits, when compared to the plan of benefits at an earlier period. In most cases, such reductions were anticipated under the circumstances that came to prevail (high health care cost increases), but were not taken into account by most of the projection and discounting methods of the time.

Current actuarial and accounting methods generate present values for terminable retiree health plans that have little credibility as measures of the beneficiary's asset or the sponsor's liability. Improvements are needed that will expand the actuarial toolbox and provide solutions to economic and accounting problems.

This paper provides a basis for discussion of assumptions that are appropriate when the plan sponsor can unilaterally and dramatically change future cash flows. The paper discusses how actuaries might best approach measurement situations where further plan reductions, or outright terminations, are to be anticipated. It introduces refinements and briefly discusses how each would fit with the usual actuarial model and how differences might affect behavior. These ideas are related to financial economics and the Bader-Gold paper "Reinventing Pension Actuarial Science." Before concluding, the wider topic of discount rate selection is briefly addressed.

INTRODUCTION

The actuarial model used to measure retiree health benefits has proven to be of limited usefulness in understanding the obligations associated with the benefits. A major failing has been to disregard the financial uncertainty that is implicit in a plan sponsor's unilateral ability to change, or even terminate, the obligation to pay for the benefits. Most sponsors continue paying for health benefits to retirees, despite the termination potential, and there are good organizational reasons to do so. In many cases in the last dozen years, however, the continuing steep increase in plan costs has led to benefit reductions, nullifying the implied promise to maintain benefits at the previous levels. In those cases, the previous actuarial measurement, which valued the changed plan many years into the future, is seen to have overstated the obligation.

If this happens very often (and it has), the measurement loses credibility. Measurement is an empty exercise if it regards as unchangeable certain contingencies that practical experience shows to be changeable in a significant way. No contingency is more significant than whether the plan sponsor will support the plan financially. When the retiree plan is terminable at the decision of the plan sponsor, measurement calls for an actuarial model that recognizes the risk of termination. Greater accuracy of measurement would reflect this risk and could improve decision making in retiree health situations.

The forum on pension actuarial practice in light of financial economics provided an opportunity to consider the retiree health situation anew. The actuarial model used for measuring retiree health obligations derives almost entirely from the pension model that was in existence 20 and 25 years ago when retiree health benefits were first brought to the attention of actuaries. If pension actuarial science is being reinvented, let's not leave behind retiree health and its actuarial arts and sciences.

The paper also addresses a larger question for actuarial science and financial economics: whether there might be a quantification model that can distinguish between a promise that is legally binding and a promise that depends on one party's willingness to pay. These are two very different promises, although there is some possibility they will be completed in identical fashion. That possibility is likely to diminish, however, as the time lengthens and circumstances intervene between the promise and its due date. The history of retiree health benefits makes this clear, so quantification concepts suggested here that work for retiree health might extend to other financial areas.

This paper examines a number of ways the actuarial model can be refined to take into account the plan sponsor's legal right to terminate a retiree health plan. After discussing the current model and its drawbacks, this paper introduces three of those refinements and briefly discusses how each would fit with the usual actuarial model. One modification leans more heavily on financial economics than the others, but in the sense that each attempts to quantify a risk previously not quantified, each may be of interest to this forum. The paper then turns to how the measurement results of these refined models would compare with those of the current model, the likely effect on behavior of interested parties if the refined model were to be adopted, and likely objections to adopting the refinements. The paper comments on some aspects of The Great Pension Controversy. The applicability of such refinements to other fields and the advantages of a more flexible actuarial model are mentioned in conclusion.