The challenge

A client with a funded postretirement health plan wanted to review the program's asset allocation to determine what the optimal risk/return strategy would be to maintain a reasonable funded status for the plan. From a fiduciary standpoint, the primary objective was to attain a well-funded program. From a practical standpoint, serious consideration needed to be given to the potential of overfunding the program, with the management of Financial Accounting Standards Board (FASB) expense being a secondary consideration.

This non-union program was funded primarily through a 501(c)(9) trust—a Voluntary Employee Beneficiary Association (VEBA). Unlike pension plans, retiree medical programs do not have to be funded, and more importantly, are restricted in their ability to fully fund due to a prohibition on funding for future medical inflation. However, also unlike a pension plan, a plan sponsor is not required to use VEBA assets to pay benefit payments. In this case, the plan sponsor elected not to use VEBA assets to pay benefit payments, and as a result the plan became quite well funded.

The challenge for the client was to fund but not overfund the retiree medical VEBA, as amounts contributed to a VEBA can only be used to provide benefits to participants, which if overfunded would potentially mean giving away benefits in excess of those provided under the plan. It is important to note that the retiree health program was available to only a closed group of current and potential future retirees, and that the expected VEBA payments would begin to decrease significantly after a certain point in time.

As a result, the plan sponsor was faced with the challenge of how to optimize investment returns, minimize accounting expense, and minimize the potential of overfunding the plan that would subject the company to "giving away benefits" as well as a potentially unfavorable tax situation.

Milliman proposed using the stochastic modeling tool, FutureCost, to examine various asset-allocation strategies, consider the variability on accounting results and impact to trust solvency expected with each reviewed allocation. As well, Milliman proposed a review of the timing of the payment of benefits from VEBA assets and the sensitivity to funded status and expense.

The solution

Milliman and the client agreed to initially evaluate three liability-efficient asset mixes based on 2011 capital market assumptions to project the solvency of the trust. The mixes were as follows:

• 80% equity/20% fixed
• 50% equity/50% fixed
• 20% equity/80% fixed

The modeling reflected two key variables to illustrate sensitivity. Benefit payments were projected under both the current trend assumption and two alternative trend assumptions (one higher and one lower than the current trend assumption) to illustrate sensitivity of results to different levels of medical inflation.

It was discovered that the timing of when benefit payments were made exclusively from the VEBA had a major impact on the results. In many scenarios, the trust assets were projected to outlast the expected benefit payments unless payments from the trust commenced within the next few years. As a result, the stochastic projections were done using two benefit payment assumptions:

• With benefits paid from the Trust beginning with the 2011 plan year and no future employer contributions.

and

• With benefits paid from the Trust beginning with the plan year when the Trust assets covered 90% of plan obligations (measured on an economic rather than accounting basis with a discount rate equal to the expected return on assets) and no future employer contributions. The 90% funded basis was used as a trigger to minimize the potential for excess assets remaining in the Trust after all obligations were met.

For both sets of payment assumptions, Milliman ran 1,000 stochastic scenarios for inflation and asset class rates of return under each of three medical trend assumptions. The projections for accounting expense and solvency focused on a ten-year horizon. Liabilities for accounting were measured on an Accumulated Postretirement Benefit Obligation (APBO) basis, and liabilities for solvency were measured on an Expected Postretirement Benefit Obligation (EPBO) basis, which provided the true economic value of plan payments.

The outcome

The initial results provided some key information to the client. If benefit payments were assumed to begin in 2011, in most scenarios the VEBA balance declines and the plan's funded status erodes. The analysis focused on projections assuming payments from the Trust when the Trust attains a 90% funded status.

These results showed the changes resulting from medical trend variation, but interestingly, variation in trend did not have as significant an impact on the funded status as investment mix or timing of payments.

In general, the scenarios utilizing the 80% equity/20% fixed were the most volatile, resulting in the highest probability of both underfunding and overfunding, while the scenarios with lower equity exposure had less risk of both overfunding and underfunding. The fatter tail risk associated with portfolios containing a higher percentage of equity did not meet the fiduciary objectives of the sponsor.

On the other hand, with benefits assumed to be paid from the Trust once 90% funded, the more conservative asset mixes projected sponsor payments from corporate assets for longer periods of time than riskier mixes (those portfolios with more equity exposure). Discussions with the sponsor led to the consideration of two additional asset mix scenarios: 30% equity/70% fixed and 40% equity/60% fixed.

The chart below shows the distribution of the cumulative ten-year APBO funded ratio for all five asset mixes considered.

Through careful consideration, a compromise between fiduciary and expense management was reached. A 35% equity/65% fixed investment mix was recommended and approved by the decision-making committee. The understanding of the balance between adequate and "too much" funding was understood by the committee, and they have agreed to revisit the study in a few years to potentially fine-tune both the asset allocation and the cash flow picture.

This case study is just one example of how using FutureCost capabilities to perform an asset-liability study for a postretirement medical plan can enable sponsors to select investment allocations that offer the best return scenario within an appropriate risk tolerance.