In the shift to liability-driven investing (LDI), corporate pension plans made progress when they changed their policy benchmarks from asset-centric total return indices to pension liabilities, as they sought to match assets to liabilities. Along with the policy benchmark change, investment objectives changed as well: defeasance of pension liabilities, not maximizing total return of pension assets, became the ultimate objective for most corporate plan sponsors. Minimizing funded status volatility became imperative, as plans deemed "at-risk" (those below 80% funded status) by the Pension Protection Act of 2006 required additional cash contributions and faced harsher benefit restrictions.
However, in the LDI strategies embraced by U.S. corporations to de-risk their pension plans, there still remain blind spots.
First, even though minimizing funded status volatility and avoiding an "at-risk" designation are imperative for most corporate plan sponsors, funded status volatility is a residual of the process, and not central to how LDI strategies are implemented and managed today. Funded status volatility fluctuates from one plan measurement date to the next, representing a major risk not actively managed by corporate plan sponsors.
Second, glide paths are simple, naive and rules-based, often based on a single statistic – the funded ratio. Such a glide path completely ignores the varying risk tolerance levels of different pension plans and ignores the risk environment in which assets are rebalanced between return-generating and liability-hedging assets. Plan sponsors sacrifice a great deal for this simplicity.
In what follows, we analyze these two deficiencies of LDI implementation, and propose (1) a better way to actively manage and minimize funded status volatility and (2) an adaptive glide path not solely dependent on the level of funded status or interest rates.
Given that pension liabilities represent the policy benchmark, the associated duration represents the single-most important risk that must be managed in any successful LDI program. As a result, the focus for most LDI programs has been on the glide path (i.e., the proper mix between liability-hedging and return-generating assets based on discrete levels of funded ratios or, sometimes, interest rates), duration matching, and liability-hedge ratios. Return-generating assets have played second fiddle to liability-hedging assets.
Exhibit 1 illustrates the breakdown of funded status volatility for a stylized 80% funded corporate DB plan with a 60/40 mix between return-generating and liability-hedging assets.
From 2001 to 2015, the estimated funded status volatility for this hypothetical plan, as shown in Exhibit 2, was about 9%, with return-generating assets accounting for 85% of the risk as demonstrated in Exhibit 1.
Some struggle with the meaning of funded status volatility. What exactly does 9% represent? Is it too high, too low? How does one relate funded status volatility to a plan's funded ratio?
Think of funded status volatility as the tracking error of excess returns between plan assets and plan liabilities (the policy benchmark). In long-only equities, only the most aggressive equity strategies exhibit that level of tracking error. Against that backdrop, a plan with 9% funded status volatility is really aggressive. In negative two-standard-deviation tail events, where risk assets fall sharply in concert and plan liabilities rise due to falling discount rates, the return from plan assets would lag the return from plan liabilities by about 18%, and the funded ratio would drop to 66% for an 80% funded plan. In negative three-standard-deviation tail events, the funded ratio would drop to 58%. No plan can afford a 14% or, even worse, a 22% drop in funded ratio in one measurement period.
To demonstrate that a drop of 14% or 22% in funded ratio is not just a theoretical possibility, in Exhibit 3, we use common asset proxies to simulate the funded ratio of the 80% hypothetical plan.
Between 2001 and 2015, this hypothetical plan experienced negative three-standard-deviation events twice and negative two-standard-deviation events four times, even though 40% of the plan assets were invested in liability-hedging, high-quality, long-duration bonds.
Clearly, 9% funded status volatility is too high. It still remains: what then is the appropriate level of funded status volatility for U.S. plan sponsors? Unfortunately, there is no prophylactic level of funded status volatility that is appropriate for all U.S. plan sponsors. Lower is better, but how much lower? That answer lies with the individual plan sponsors.
For any corporate defined benefit plan, funded status volatility is too great a risk not to manage actively. Even though pension liabilities represent long-term obligations, corporate plan sponsors do not have the luxury to ride out funded status volatility because pension regulations require annual measurements of pension assets, pension liabilities, and funded status, and there are negative consequences if the drop in funded ratios coincides with one or more measurement dates.
"Plan sponsors must think long term when it comes to investment holding periods, but short term when it comes to managing funded status volatility."
Suny Park, CFA, CPA
Much of what we have covered thus far is well understood by most corporate plan sponsors. Most LDI practitioners recognize the time-varying nature of funded status volatility and understand that between 80% and 90% of funded status volatility emanates from return-generating assets. However, most corporate plan sponsors take a passive approach to managing the asset allocation within return-generating assets.
The current LDI practice is inconsistent and incomplete with respect to minimizing funded status volatility. We assert that corporate plan sponsors can more effectively minimize funded status volatility by actively managing beta exposures within their return-generating assets.
Since strategic asset allocation (i.e., passive beta management) within return-generating assets is the source of most of the funded status volatility, the minimization of the latter often requires active, not passive, beta management. By active beta management, we are not advocating market timing or forecasting asset class returns, rather, an adaptive asset allocation approach based on forward-looking measures of tail risk (both downside and upside) derived from the option markets.1
Active beta management may sound like an anathema to most institutional investors. After all, it is foolhardy to make interest rate and equity market-timing calls. We wholeheartedly agree. No one is clairvoyant enough to consistently time interest rates and equity markets. Therefore, in active beta management, we advocate for neither market-timing, nor forecasting asset class returns.
We advocate for a fundamentally different approach to asset allocation: adaptive, as opposed to static, asset allocation based on forward-looking measures of risk (both downside and upside) not dependent on valuation levels, factor timing, or market-timing calls. We are careful to distinguish between risk (how much money one can lose) and volatility (the variability of returns).2 In conventional strategic asset allocation, one takes long-term capital market forecasts of returns, volatilities and ordinary correlations as inputs to derive passive dollar allocations to each of the strategic assets. In the adaptive asset allocation approach we propose, one targets a level of portfolio risk (for example, maximum loss of 25% over a 12-month period), and estimates short-term, forward-looking measures of risks and tail-risk correlations3 as inputs to determine asset allocation that adapts to the changes in the beta risk environment. It adapts to maintain the downside risk of the portfolio consistent with the maximum loss target.
However, active beta management based on short-term changes in downside risk represents only one side of the equation. To maximize compound return from active beta management, it is not enough to dynamically change asset allocation based solely on downside risk; one must also dynamically adapt the asset allocation to capture the upside or the right tail risk.
Option markets offer a forward-looking measure of risk (both downside and upside) that is not dependent on valuation levels, factor timing, or market-timing calls. Further, such a risk measure dynamically adapts to the changes in the risk environment: when the probability of loss for the underlying asset increases, the price of the put option increases; similarly, when the probability of gain for the underlying asset increases, the price of the call option increases.
Therefore, we believe option market-implied, forward-looking measures of downside and upside risk are an ideal way to actively manage beta exposures: increasing allocation to assets with low expected losses and high expected gain-to-loss ratios, and decreasing allocation to assets with high expected losses and low gain-to-loss ratios.
To be clear, we are not advocating for the buying and selling of options to manage beta exposures; rather, deducing forward-looking measures of downside and upside risk from the price of options on equities, credit, global sovereign bonds and inflation-sensitive assets.
In Exhibits 4a and 4b, we demonstrate the benefits of an adaptive asset allocation approach within return-generating assets.4
Exhibit 4a shows the improvement in the stylized pension plan's funded status as the adaptive asset allocation portfolio is increased to 25%, 50% and 100% of return-generating assets. To be clear, a 25% allocation to the adaptive approach within return-generating assets represents only 15% of the total plans assets (25% of 60%). But the benefit was substantial: the ending funded status increased from 72% to 81% to 91% to 114% in concert with the increases to the adaptive asset allocation portfolio.
Exhibit 4b demonstrates the decreases in the three-year rolling funded status volatility. Over the entire analysis period, the stylized plan's funded status volatility decreased from 8.8% to 6.5%, based on the increases to the adaptive asset allocation portfolio within return-generating assets. There are three notable observations:
Two key insights are worth calling out: first, plan sponsors cannot lower the level of funded status volatility while simultaneously maintaining a static mix of risk assets in return-generating assets. In fact, in our analysis, their ability to lower funded status volatility was directly related to the level of adaptive beta management within return-generating assets. Second, to minimize the variability of funded status volatility, plan sponsors cannot maintain a static mix of return-generating and liability-hedging assets that current LDI glide paths call for.
In an ideal world, corporate plan sponsors would target a constant level of funded status volatility in their LDI programs. A plan – with a constant level of funded status volatility that imparts some semblance of constancy and predictability – would be highly desirable for most pension officers and CFOs in their financial planning and analysis efforts.
However, as we noted in the previous section, plan sponsors simply cannot expect a constant level of funded status volatility if they maintain a static mix between return-generating and liability-hedging assets. As our example indicated, even when the adaptive asset allocation portfolio was included within return-generating assets, funded status volatility remained variable because the balance between return-generating and liability-hedging assets remained static at 60/40. To achieve a constant level of funded status volatility, plan sponsors should consider an adaptive glide path that responds to the changes in risk environment.
By employing an adaptive asset allocation approach at the overall plan level, plan sponsors could completely do away with the naive glide path. They would:
Unfortunately, an LDI program that targets a constant level of funded status volatility will exist only in an ideal world. Although it is intellectually stimulating to explore, it is too revolutionary and disruptive for most, especially for those who have already embarked on an LDI strategy. It introduces a level of operational complexity that no plan sponsor wants.
As a compromise, we propose an adaptive glide path that can improve on the naive glide path, which is solely dependent on the level of funded status. This does not eliminate the existing glide path altogether, but seeks to add more intelligence and flexibility to it.
At the beginning of November 2016, the normalized, tail-based Sharpe ratios for U.S. equities and U.S. 10-year Treasurys were 0.9 (or about negative one-standard deviation away from the historical mean) and 0.8 (or about negative two-standard deviations away from the historical mean), as shown in Exhibit 6.
Consider a hypothetical plan whose funded ratio improved from 75% to 80% simply due to an additional cash contribution from the company. Without the rebalancing range presented in Exhibit 5, the naive glide path would force a plan sponsor to automatically change the mix between return-generating and liability-hedging assets from 60/40 to 50/50. The additional cash contribution would have to be invested consistent with this new asset mix, despite the fact that the option market-implied tail-based Sharpe ratio for 10-year U.S. Treasurys currently stands at more than negative two-standard deviations away from the historical norm.
In the absence of forward-looking measures of risk, a simple, rules-based glide path and the resulting asset allocation decision may be acceptable. However, when armed with information about the potential downside risk (relatively high) versus potential upside gain (relatively low) for 10-year Treasurys, one should pause before increasing the allocation to liability-hedging assets from 40% to 50%. The sensible thing to do would be to maintain the existing asset mix of 60% return-generating and 40% liability-hedging assets – consistent with the rebalancing ranges – and hold the additional contribution from the company in cash until the ETG to ETL ratios for both U.S. equities and 10-year Treasurys normalized.
Stepping back and objectively assessing the LDI glide path strategy currently implemented across most U.S. corporate defined benefit plans, one cannot escape the fact that it is naive and entirely rules-based. When the funded ratio improves from 75% to 80%, the glide path may call for a shift in asset allocation – from a 60/40 mix to a 50/50 mix between return-generating and liability-hedging assets without any regard to the existing risk environment. The beauty of this approach lies in its simplicity.
Unfortunately, plan sponsors sacrifice a great deal for this simplicity. The naive glide path requires them to be completely risk agnostic: if the funded status improves from 75% to 80%, the plan sponsor must increase the allocation to liability-hedging assets even if the expected tail loss associated with bonds (i.e., duration risk) is high or increasing. If the funded status deteriorates from 80% to 70%, this naive glide path may require plan sponsors to increase the allocation to return-generating assets even if the expected tail loss on equities is high or increasing. In the former case, the plan sponsor may rebalance to long-duration fixed income portfolios when interest rates are rising; in the latter, the plan sponsor may rebalance to equities when equities have further to fall – in essence catching a falling knife. This is precisely the wrong thing to do when one wishes to maximize compound returns and improve the overall funded status of the plan.
In the current environment, where returns are hard to come by and the drumbeat of rising interest rates is getting louder and louder, following a naive glide path solely based on funded status without any regard for tail risks in return-generating and liability-hedging assets may lead plan sponsors down a suboptimal path. To improve funded status, to lower funded status volatility, and to minimize ongoing cash contributions to their respective defined benefit plans, it behooves plan sponsors to challenge the status quo and take a fresh look at the design of their LDI glide paths and the construction and management of their return-generating assets. A blind glide path that ignores risk – more specifically, tail risks – and static asset allocation, or passive beta management, within return-generating assets run counter to corporate plan sponsors' desires to improve the funded status and minimize the funded status volatility of their pension plans.
Investment Insight Author
Published January 2017