Matching the duration of fixed-income investments to those of a plan's liabilities is common practice to hedge the interest rate risk in pension plan funding levels. However, in an effort to close their funding gaps, plan sponsors maintain significant allocations to equities and other return-generating assets because of higher expected returns. As pension plans become more cautious about return expectations, we examine the impact of incorporating non-fixed-income assets in LDI allocation policy and manager selection due diligence. We demonstrate that this exercise has provided a low-cost "insurance policy" against funding level declines when a plan's investment portfolio underperforms.
In typical liability-driven investment implementations, the return-generating portion of the portfolio is not considered when evaluating the overall portfolio's asset-liability-matching relationship and interest rate sensitivity. Ignoring the long-term, albeit weak, relationship between these return-generating assets and interest rates can lead to suboptimal results: Each derisking milestone affords a plan fewer opportunities to reach subsequent derisking milestones. Acknowledging the contribution of non-fixed-income assets to funding-level volatility enables plans to maintain higher allocations to return-generating assets than in a traditional LDI framework. Incorporating the right kinds of non-fixed-income assets in an LDI implementation allows plans to participate more in potential market upside while protecting funding level when the non-fixed-income assets underperform.
We refer to portfolio duration as the historical tendency of a pension plan's overall investment portfolio to match the change in the plan's liabilities. More specifically, portfolio duration is defined as the absolute difference between one and the trailing five-year beta of a pension plan's overall investment portfolio returns to the change in the plan's liabilities.
The lower the portfolio duration, the better the pension plan's overall portfolio has acted as a hedge against its liabilities in the past. While market risks shouldn't be thought of as a hedge for liabilities, risks that have been correlated with liabilities in the past have provided a tactical opportunity for pension plans to allocate to return-generating assets.
While investment return, starting funding level and change in inflation are key drivers of change in funding level, portfolio duration, when including return-generating assets, is an incrementally significant driver, above and beyond the role of the change in 10-year Treasury yields. We regressed the year-over-year percentage change in funding level against overall portfolio return, percentage change in inflation, percentage change in the 10-year Treasury yield, and portfolio duration using investment performance and liabilities data for 173 pension plans from 2001 to 2016 (Table 1).
The results indicate that a one-unit decrease in portfolio duration is estimated to increase the average pension plan's funding level by 0.06% of its starting level, after controlling for the other drivers previously mentioned. That's probably not enough to get anyone excited. When a pension plan's investment portfolio produces positive returns, portfolio duration really doesn't matter. But when a pension plan's investment portfolio produces negative returns, the importance of portfolio duration as an actionable insurance policy starts to becomes evident. Regressing the year-over-year percentage change in funding level against the same drivers for the subset of the same data in which the plan's overall investment portfolio produced negative returns begins to tell a different story (Table 2).
In periods when a plan's investment portfolio produces negative returns, a one-unit decrease in portfolio duration is estimated to increase the average plan's funding level by 0.25% of its starting level, after controlling for the other drivers. In dollar terms, a one-unit decrease in portfolio duration is estimated to be worth an average of $25 million for every $10 billion in plan assets in down years, holding liabilities and all else constant.
To further illustrate, we split the 16-year sample of pension plans into two parts based on the median pension plan annual return of 8.44% and define the subset of the data in which plans returned less than the median as the periods in which the portfolio underperforms. For the periods in which pension plan portfolios underperformed, we then split the pension plans into 10 buckets based on portfolio duration and plot each bucket's median change in funding level (Figure 1). The median pension plan in the decile with the lowest portfolio duration saved 2 percentage points of funding level over the median pension plan in the decile with the highest portfolio duration.
The power of portfolio duration holds up even during the 2008 global financial crisis (Figure 2) — pension plans that maintained lower portfolio durations have been rewarded, on average, with superior funding level protection than the rest of their peers. The median pension plan in the quintile with the lowest portfolio duration saved 9 percentage points of funding level over the median pension plan in the decile with the highest portfolio duration.
Maintaining allocations to return-generating assets doesn't have to be as much of a trade-off if the risks are chosen to more closely match the liabilities. Such an approach within an LDI framework can minimize liability tracking error, which contributes to quicker achievement of funding milestones, especially during periods in which the overall portfolio underperforms. We suggest plan sponsors should replicate this analysis using their internal investment and liabilities data to evaluate whether portfolio duration could fit in their allocation policy and manager selection decision making process.
Nishang Gupta is product analytics architect at Thornburg Investment Management Inc., Santa Fe, N.M. The views expressed are subject to change and do not necessarily reflect the views of Thornburg Investment Management. This content represents the views of the author. It was submitted and edited under P&I guidelines but is not a product of P&I's editorial team.