Assisting a plan sponsor with its investment portfolio using asset liability modeling

  • Print
  • Connect
  • Email
  • Facebook
  • Twitter
  • LinkedIn
  • Google+
Door Zorast Wadia | 12 december 2018

The challenge

A publically owned client in the financial services sector had just emerged from a divestiture from its parent company. The defined benefit pension plan’s investment allocation was entirely dedicated to fixed income instruments and cash given the need for portfolio stability during the divestiture. The plan sponsor asked Milliman to assist with the subsequent re-risking of the plan’s investment portfolio. While the pension plan was currently funded on an ERISA basis so that there was no minimum required contribution due, it was underfunded on an accounting basis and also subject to Pension Benefit Guaranty Corporation (PBGC) variable rate premiums as a result of not being fully funded on a PBGC basis.

The plan sponsor wanted to explore investment options that would best allow it to meet its future contribution requirements while limiting the risk of being underfunded on either an accounting or PBGC basis. The plan sponsor also wanted to understand the impact on accounting expense and balance sheet volatility as a result of investment allocation changes where the employer would take on more equity exposure.

The pension plan had a legacy/grandfathered pension formula (based on final average pay and service) that was closed to new entrants and a cash balance formula in effect for participants hired after 2016. Grandfathered participant benefits would continue to accrue and the cash balance benefits were expected to grow as well over the next decade. Thus, both components had to be uniquely considered in any long-term planning scenario. Furthermore, while the grandfathered accrued benefit could be paid only as an annuity, the cash balance benefit was payable as a lump sum upon vested termination. Therefore, the liquidity needs of the investment portfolio for the pension plan also had to be considered. The pension plan had assets of roughly $1.5 billion at the beginning of 2018 and was 112% funded on an ERISA basis and 91% on accounting basis.

Milliman solution

Milliman consultants worked with the client to review its investment policy statement, understand the risk tolerance, set achievable financial goals, and present projections of assets and liabilities as part of an end deliverable in the form of an asset liability modeling (ALM) study. The goal of the ALM study was to estimate expected levels, trends, and possible variability over the next 10 years of the plan’s annual required contributions, funded status, and accounting expense under the current policy asset allocation and several alternative asset allocations based on the client’s input. The ALM study used 10,000 stochastic projections of the growth in both plan liabilities and assets over a 10-year period based on consensus capital market assumptions developed by Milliman investment actuaries along with input from the plan sponsor. Milliman pension actuaries developed valuation and projection assumptions including parameters to determine new entrant profiles based on the plan’s existing demographic experience.

Milliman reviewed the investment policy statement to make sure the alternative investment portfolios being modeled were consistent with the governing plan documentation. After several discussions, Milliman and the client together determined the following financial objectives and used them as parameters in the ALM study:

  • Produce a more stable and predictable cash contribution pattern
  • Avoid PBGC variable rate premiums
  • Improve accounting funded status through increased returns based on alternative asset allocations

Milliman employed Monte Carlo forecasting techniques to project assets and liabilities. Milliman’s ALM study consisted of two main stages. In the first stage, the risk budget was set so that the split between equities and fixed income investments was quantified. The structure of the fixed income investments, including the degree to which the portfolio would be duration matched against plan liabilities, was also explored as part of the stage one analysis. Upon receiving additional feedback from the plan sponsor, the goal of the second stage of the ALM study was be to allocate the risk budget specified from stage one via equity diversification within the portfolio.

The ALM study used several key metrics to quantify risk levels for the pension plan. These included contribution cost and volatility, the expected long-term growth rate of assets, funding and accounting funded ratios, tracking error, balance sheet risk, and accounting cost and volatility. Milliman summarized the year by year projection results for contributions, PBGC premiums, funded status, and accounting expense using tables and floating bar charts.

The outcomes from the 10,000 stochastic scenarios were ranked into percentiles to communicate the likelihood associated with each of the summarized results. As a contrast, under a deterministic approach, single point estimates would be provided but the risk associated with the given results would not be quantified. As a result of this important difference, the stochastic ALM analysis was deemed to be the more appropriate deliverable.

The outcome

Milliman prepared a comprehensive ALM report at the conclusion of the stage one analysis. This report was reviewed with the plan sponsor’s investment committee and the risk budget was finalized. Subsequently, recommendations for specific alternative portfolios to be modeled in the second stage of the ALM analysis were obtained. After re-running the stochastic projections a second time based on the refinements provided by the client, a final ALM report was produced and delivered to the client for review.

The final ALM report was truly a collaborative process between the plan sponsor and Milliman investment and pension actuaries. Risk and reward considerations were discussed with the client when presenting the outcomes from the initial ALM analysis. The client’s final decision on portfolio allocation changes would be dependent on its risk tolerance and the extent to which it would accept the results shown in the study. The client was pleased with Milliman’s consultative approach and is currently reviewing several of the portfolio alternatives modeled by Milliman.