In liability-driven investing (LDI) it is common practice for pension plans first to construct a bond portfolio that matches the risk profile of their liabilities using a predetermined fraction of their assets, and then allocate the remaining fraction to risk assets, such as equities, real estate, and private equities. Frequently, the allocation among the risk assets is performed independent of explicit consideration of the liabilities and is strongly tilted to equities.
This two-step approach might be appealing from a mental accounting perspective, but we believe it is suboptimal. The remaining unhedged portion of the liabilities interacts with risk assets in generating surplus (or funded status) risk. Therefore, plan sponsors should consider allocating their risk asset portfolios in a way that explicitly integrates liabilities, consistent with the approach they use to build their bond portfolio. In other words they should take into account correlations between risk assets and liabilities. To the extent that the unhedged portion of the liabilities represents a short duration exposure of the plan, it may be optimal to allocate to risk assets that carry positive duration exposure, such as emerging market debt and private equity, in addition to equities.
Key Definitions
Funded status: The difference between the value of the asset portfolio and the present value of the liabilities as discounted on the AA curve.
Duration factor: An asset or a portfolio’s sensitivity to parallel changes in interest rates.
Credit factor: An asset or a portfolio’s sensitivity to changes in the level of credit spreads in the market.
Standard deviation: A measurement of the dispersion of a set of data from its mean. Lower standard deviations indicate that the data is generally close to the mean, while higher standard deviation measurements indicate data points are spread out over a larger range of values.
|
|
The risk factor perspective
Asset allocation, liability modeling, and immunization analysis are non-trivial endeavors. Integrating all three requires considerable effort. Liability valuation itself, and the measurement of surplus, while exact from the perspective of accounting and actuarial standards, are only approximations of the underlying economics of meeting liability payments over the next 30 to 100 years.
A key tool in successfully integrating asset allocation, liability valuation, and immunization is the risk factor perspective. Risk factor analysis decomposes assets and liabilities into their fundamental “risk off” and “risk on” building blocks. Because they are discounted on the AA curve, liabilities are exposed to a “risk off” factor (duration, i.e. interest rate risk), as well as a “risk on” factor (credit spread risk). Of course, the duration exposure is the dominant of the two.
Diversification in times of market stress
To assess diversification across assets and liabilities requires careful analysis of correlations during times of market stress. Indeed, it has been widely observed that correlations in down markets differ from correlations estimated from the full sample of returns. Therefore, full-sample correlations can overstate – or understate – a risk factor’s diversification properties in market environments when diversification is most needed. The duration-equity and credit-equity correlations are especially sensitive to this effect.
|
Figure 1 compares how duration and credit have diversified equity risk historically, based on close to forty years of monthly data (from February 1973 to February 2012). It shows full-sample correlations, as well as tail event correlations, which are calculated on months during which equities – as represented by the total return on the Wilshire 5000 Index – were down by two standard deviations or more relative to their average return. Standard deviations are calculated over the full sample. Here, duration is represented as the total return of the Barclays Intermediate Treasuries Index, while credit is return per unit of duration for the Moody’s BAA spread over Treasuries.
Over the full sample, both correlations are low and slightly positive: 8% for duration and 3% for credit. However, during tail events, duration typically experiences a flight-to-safety rally, while credit spreads “blow up” (which means, credit crashes with equities). Figure 1 also shows the tail event correlation between duration and equities is - 48%, compared to + 82% for the credit risk factor.
From an asset-only perspective, the obvious conclusions from this correlation analysis are that: a) in a risk selloff, the portfolio may benefit from a rally in Treasuries, and b) credit is a poor diversifier of equity risk in the portfolio. But from an asset-liability perspective, these duration diversification dynamics must be interpreted carefully – most plan sponsors are short duration because their liabilities have longer duration than their total asset portfolio. If rates rise, they may benefit. But for those plan sponsors, there’s no such thing as a flight-to-safety diversification effect between duration and equities. During a “risk off” event, they will likely experience a double whammy: their equity portfolio will likely lose value and their liabilities will balloon faster than their bond portfolio, on a dollar value basis. Hence, plan sponsors who wish to reduce their funded status risk should consider increasing allocation to Treasuries and recognize their funded status may benefit from a rally in duration during equity drawdowns only once they become net long duration. Plan sponsors who are unwilling to increase duration or face constraints in this regard should pay special attention to identifying alternative sources of diversification within their risk assets.
Overall, when diversifying across risk assets, there are choices that may be more attractive to pension plans than they are to liability-agnostic investors, such as risk assets with exposure to duration. The exposure may be explicit fixed income exposure, of which emerging market debt would be a clear example, or exposure to private and illiquid asset classes whose valuations are strongly linked to present value discounting processes.
Solution: a diversified portfolio of alternatives
Putting it all together, plan sponsors who choose to maintain a short duration stance on a total portfolio basis should consider alternative sources of diversification beyond equities, and they should focus on risk assets with indirect duration and credit exposures.
Figure 2 shows an example of the diversification benefits that may be achieved by substituting a portion of a plan’s equity portfolio for commodities, emerging market local debt and an absolute return strategy. The equity focused allocation is comprised of 70% equities, while the diversified risk assets allocation substitutes half of the equity allocation for an equally-weighted portfolio of these three alternative asset classes. Both allocations invest 30% in an LDI immunization, fixed income portfolio. The bottom of the exhibit compares risk decompositions for the assets themselves, and for the assets net of the liabilities (“surplus” risk decomposition). Decomposing surplus volatility into its risk factor contributions reveals that while duration potentially reduces portfolio risk on an asset-only basis, when the pension plan is net short duration, it may contribute to portfolio risk on an asset-liability basis. However, this does not mean that exposure to assets with duration is contributing to risk: rather, it is the insufficient exposure to duration that causes the risk.
This solution offers 3% to 4% in potential volatility reduction. The cost of this risk reduction depends on the difference between assumed returns for equities and for the diversified risk asset portfolios. Once assumed returns are determined, plan sponsors should assess how much return they are willing to sacrifice in order to reduce risk. While this example was not explicitly optimized, an optimization process that incorporates the liabilities can be used in an effort to efficiently scale the risk asset positions. This example is meant for illustrative purposes, as different portfolio structures will be appropriate for different investors and yield different results.
In general, from an LDI perspective, plan sponsors who are constrained or choose to remain short duration may want to consider risk assets that offer alternative sources of equity risk diversification, such as commodities, as well as assets with stable cash flows – and hence greater indirect duration and credit exposures – such as high-dividend stocks, real estate, infrastructure and some hedge funds. Duration exposure from markets that are not at their zero-bound and for which debt-to-GDP levels are reasonable, such as emerging market bonds, may also provide incremental returns while reducing risk to the funded status. In all cases, plan sponsors should carefully evaluate correlations in times of stress before optimizing their portfolio.
Takeaway: Alternative sources of diversification
In liability-driven investing, unless the plan is fully immunized (fully invested in a liability-matching portfolio) or significant leverage is employed, the bond portfolio only hedges part of the liabilities. Hence, plan sponsors should consider applying the same concepts they use to construct their LDI bond portfolio to the selection of risk assets. Doing so reveals the need to balance risk assets across alternative asset classes differently than from the asset-only perspective, especially as it relates to the role of direct and indirect credit and duration exposures in the portfolio. Investors who must remain net short duration should consider alternatives sources of diversification as well as indirect duration and credit in their risk asset portfolios. Several asset classes offer such characteristics, such as emerging markets bonds, commodities and hedge funds.
Rene Martel, Vijendra Nambiar, Fei He and Jared Gross contributed to this analysis.