A version of this article has been published in Institutional Investor’s Journal of Retirement. For more information please visit www.iijor.com.
As the global shift away from traditional defined benefit pension plans continues, workers are increasingly reliant on defined contribution (DC) pension schemes to build their own retirement income. Australia, the U.S. and the U.K. lead in global DC development, with 81%, 58% and 26% of all pension assets, respectively, held in DC plans at the end of 2012 (Towers Watson Global Pension Assets Study, January 2013). DC plans will need to provide more than 60% of all retirement income for workers in these countries, according to the Organisation for Economic Co-operation and Development (OECD).
What is remarkable about the financing of retirement income in these three countries is that although retirees in each country rely heavily on DC plans, plan design and investment management differ dramatically. There is no obvious rationale for these differences, given the basic similarities in the institutions of retirement income finance. These similarities argue for the application of common principles of investment strategy, which could enhance the role of DC plans in the provision of a secure retirement. Of course, an adequate saving rate is a necessary condition for a financially secure retirement. However, investment design, and especially the design of default strategies, is also vital to help build and preserve purchasing power and minimize volatility and market-shock risk. It is critically important to recognize that most workers can participate in the labor market for only so many years and save only so much of their income. Many workers may be living healthier, longer lives in all three countries. Nonetheless, many workers would find extending their working life beyond the norm very difficult, and an increase in saving late in a career will have only a modest impact on the retirement nest egg. Given these limitations, DC investment structures should operate within a worker’s loss capacity – never risking more than a participant could afford to lose.
The thesis of this article is that the investment strategies that have been adopted for most workers in Australia, the United States and the U.K. leave them exposed to a degree of risk that is beyond their capacity to bear. We believe most DC plans in the three countries would benefit from an analysis of the risk of loss in their investment defaults, as well as an analysis of the sources of risk. Given the fact that many workers do not make an active choice of a particular asset allocation, we encourage plan sponsors, trustees and providers to adopt default solutions that focus on the plan’s desired outcome: helping participants replace a target percentage of final pay (e.g., 40 to 60 percent) necessary to maintain their standard of living in retirement. That desired outcome, or investment objective, helps define a worker’s risk capacity as well as appropriate asset allocation and should be consistent across global jurisdictions (while allowing for subtle nuances in local market requirements). The article will contrast current practices in the three countries with what we term outcome-focused solutions, which embrace risk-capacity limits and are designed for risk diversification, inflation hedging and market-shock protection in a way that should limit DC savers’ risks.
DC contributions … just do it!
Prior to the enactment of the Pension Protection Act of 2006, American employers likely wasted many millions of dollars trying to persuade workers to contribute to their DC plans. Among other reforms, the Act gave plan sponsors statutory authority to auto-enroll eligible employees into the plan (yet allow participants to opt out if they preferred), finally giving sponsors an alternative to begging workers to opt in. This “just do it” auto- enrollment approach, now adopted by 47% of U.S. employers, according to Plan Sponsor Council of America (PSCA), has successfully offset natural human inertia to improve DC participation and contribution rates, with about 80% of all eligible participants making contributions. What’s more, auto-escalation of the contribution rate, a feature utilized by almost 40% of plans that auto-enroll, may help pump up the percentage of salary that Americans contribute each year to 401(k) and other employer-provided defined contribution plans.
However, efforts to increase accumulation are being partially offset by plan leakage. In the U.S., an estimated 10% of DC assets annually flow out of plans, primarily as workers move from employer to employer, thus reducing the buildup of defined contribution savings. Although the majority of the departing assets may be retained for retirement if they are rolled over into Individual Retirement Accounts (IRAs), often costs in these retail programs are higher than DC plans and thus reduce future asset growth potential.
Australia and the U.K. have taken a more aggressive approach toward DC savings. In Australia, employers are required to contribute 9.25% of pay, rising to 12% by 2020, to a tax-advantaged retirement plan (overwhelmingly a “superannuation” DC program). Between 2012 and 2017, the U.K. is phasing in a requirement for employers to auto-enroll participants at a rate that will increase to 8% of pay with at least 3% contributed by the employer (employees may still opt out). In contrast to the U.S., once the money is in the Australian or U.K. programs, participants generally cannot withdraw funds until retirement age. Clearly, DC account values will build far more swiftly in the Australian and U.K. systems, given that their higher contribution rates and their firmer control of leakage. In addition, the opting-out approach seems to function more effectively in the U.K. where many companies report that more than 85% of members defaulted into plans do not opt out. American regulators are just beginning to study how to slow leakage from DC plans.
With the continued evolution toward a DC-based pension system in the U.S., and the increasingly disciplined approaches of Australian and U.K. contribution schemes, the case for ensuring that default strategies are robust and appropriate has never been stronger.
DC investment defaults … get the risk right
But how are assets invested when participants are auto-enrolled into the plan? The answer: the money is defaulted into asset allocation strategies in one form or another. For instance, in the U.S. last year, 73% of DC plans with automatic enrollment defaulted to target-date strategies, 11% to a series of single balanced options and another 9% to a series of target-risk portfolios, according to a 2013 survey by the PSCA. In Australia, generally the default went to a single tactical balanced fund. While in the U.K., plans defaulted to a “lifestyle” allocation – an approach similar to a U.S. target date strategy, which shifts participants to a more conservative asset mix as they approach retirement.
There are stark differences among these defaults in asset allocation and the allocation of the risk budget across asset classes, as well as in the implied risk of loss. These differences are puzzling, given that the ultimate challenge of DC contribution and investing – building retirement wealth sufficient to provide sustainable income from retirement until death – is basically the same in all three jurisdictions.
Let’s examine these plans from several perspectives: asset allocation, risk drivers, inflation hedging and probability of loss.
As Figure 1 shows, typical default asset allocations in the U.S., the U.K. and Australia all have significant weights in equities and introduce nominal bond exposure at varying weights and times. The traditional U.K. approach encompasses a particularly aggressive allocation to equities up until close to retirement age at which time the glide path shifts dramatically to fixed income. Although more modern approaches being advocated include a significant allocation to “
ersified growth portfolios” as an equity substitute, equity remains by far the dominant risk borne by U.K. participants for much of their savings journey. The U.S. and Australia include inflation-related assets, while the U.K. tends not to, although many of these diversified growth portfolios tend to reference a specific real rate of return objective (e.g., CPI-plus 3%). Australian balanced allocations stand out for including alternatives such as private equity, real estate and hedge funds. By comparison, the outcome-focused glide path in the U.S. includes more inflation-hedging assets and reduced allocations to risky assets over time relative to the other three asset allocation approaches.
We believe it’s important to consider not only the asset allocation, but also the overall volatility that a participant may face, as well as what may drive this volatility. Managing volatility is especially important when considering a default strategy’s risk profile relative to a participant’s ability to accept risk – risk capacity – or stomach risk – risk tolerance. The U.S., U.K. and Australian DC plans all allow participants to change their asset allocations. Thus, if the default’s risk profile exceeds the participant’s risk capacity or tolerance, people may move their allocations to lower-risk assets when markets decline, lock in losses, and thus potentially jeopardize their ability to build a retirement portfolio sufficient to fund a retirement-income investment horizon that may be decades long.
In the U.S., for instance, we know that the movement out of equities and into cash within target-date mutual funds correlates with the fluctuations of the S&P 500, particularly in down markets and most strongly for participants closest to retirement (PIMCO (2012)). Managing DC investment defaults in an effort to diversify sources of risk and minimize overall volatility may help DC participants stay the course. Not surprisingly, given the defaults shown in Figure 1, the dominant source of market risk for participants furthest from retirement comes from equities (see Figure 2). Even advances in the U.K. market to encourage investments in “diversified growth portfolios” still leave participants singularly exposed to world equity risk. This means workers’ assets will be highly sensitive to movement in stock markets.
Unfortunately, when we consider asset allocation for participants at retirement, heavy equity risk remains in the U.S. and Australia, but a complete shift to fixed income risk occurs in the U.K. approach – a reflection of the common practice of U.K. participants to purchase an annuity immediately upon retirement (see Figure 3). We believe the heavy equity risk, particularly at retirement, may promote self-destructive behavior. For example, a substantial market correction in the years leading up to retirement might prompt an excessive flight to cash.
Another notable difference among the asset allocations of the three countries is the role (or lack thereof) of inflation-hedging assets. We believe the inclusion of assets that are either indexed to inflation or with rates of growth that are highly correlated with inflation such as inflation-linked bonds (e.g., Treasury Inflation-Protected Securities), commodities and real estate is essential, as these investments best align with the DC investor’s objective – that is, for income to keep pace with the cost of goods and services needed in retirement. Inflation hedging may shift from more volatile assets early in one’s career to inflation-linked bonds, which tend to be less volatile, as retirement approaches.
Figure 4 compares the three countries’ glide paths, which show little or no exposure in the U.K. and light allocations in both the typical U.S. and Australian asset allocations. Notice that the outcome-focused glide path suggests both a higher allocation and a shifting allocation to potentially lower volatility inflation-hedging assets.
Probability of loss
Finally, given these asset allocation structures, what is the probability of loss at different times prior to retirement—e.g. 5, 10 or 15 years? We believe this may be the most important question for plan sponsors to pose. Understanding and managing volatility is important from a risk tolerance perspective, while quantifying and limiting the risk of loss is vital from a risk capacity standpoint. An investor’s risk capacity may be defined as the amount an investor can afford to lose in their account without materially jeopardizing their ability to meet their retirement income goals, including the timing of retirement and retirement lifestyle. If we consider the value at risk using a 95% confidence level, for the at-retirement allocation there is a potential 12-month mark-to-market loss of 11% for the U.S. market-average, 16% for Australia and 23% for the U.K. asset allocation (the maximum loss for the U.K. is less when viewed in an asset-liability context if the investor purchases an annuity). By comparison, the outcome-focused glide path potentially reduces this risk to 7% (see Figure 5). Many plan sponsors may want to evaluate the risk of loss at a higher confidence level, such as 99%, to consider the possible market shocks, such as the 2008–2009 financial crises.
To seek further risk reduction, and in particular to mitigate the risk entailed by extreme market conditions, a tail-risk hedging strategy may be added to asset allocation portfolios. For example, the adoption of defensive option strategies, such as put options on the S&P 500 that increase in value when the market declines, could limit losses on the long-term performance of asset allocation strategies. Figure 6 shows the potential impact of adding tail-risk hedging to the U.S. market average and, separately, the outcome-focused glide path. At the 95% confidence level, the value at risk declines by up to 20%, with the risk of a 12-month loss falling to 8% for the market-average glide path and 6% for the outcome-focused glide path at retirement. Tail-risk hedging could similarly be tailored to the risk profiles of the U.K. glide path and the Australian balanced allocation to help guard retirees against significant market shocks.
Distribution of outcomes
Putting this together, to what extent may these differences across the glide paths contribute or detract from the long-term performance of a participant’s portfolio and, more importantly, their ability to retire as planned? Figure 7 shows a graph and chart of the distribution of potential outcomes across the different glide paths using capital market assumptions in a normal market environment; this analysis also includes a glide path based on the U.S. market average but enhanced by adding a simple tail-risk hedging strategy. The reader will observe that the outcome-focused glide path with the height of the distribution and shortness of its tails offers the tightest distribution of estimated income replacement (i.e., the highest certainty of outcome) with median income replacement at 58% and 27% at the first percentile of trials. Similarly, adding tail-risk hedging to the market average glide path significantly increases the tightness of the distribution and modestly reduces the first percentile left tail outcome, yet also lowers the potential median and right tail outcomes relative to the market average glide path. This is a trade-off for both an outcome-focused glide path and glide paths which employ tail-risk hedging: increasing the potential to lower left tail outcomes and the tightness of the distribution of outcomes, while potentially lowering the upside in stronger market environments (e.g., the market average glide path with tail-risk hedging reduces the estimated income at the 95th percentile from 134% to 120%).
By maintaining a higher allocation of equities throughout the glide path, the U.S. market average and Australian glide paths offer the highest estimated median income replacement rates at 60% and 59% respectively. However, as noted earlier, holding higher risk allocations, particularly near retirement, may exceed both the participant’s risk capacity and risk tolerance, which may cause behavioral responses that would result in participants’ income replacement levels to be below what this analysis would suggest. The U.K. traditional lifestyle glide path with its lower levels of diversification and lack of tail-risk hedging shows the lowest estimated outcomes both at the median 55% and first percentile 17%, though these results may be viewed more favorably in an asset-liability context. Keep in mind that the estimates of this distribution analysis have been derived using capital market assumptions for a normal market environment. Under inflationary and turbulent markets, the inclusion of real assets, more diversified asset allocations, and tail-risk hedging strategies may better help guard the purchasing power of participants. For example, Figure 8 demonstrates the distribution of potential outcomes in which participants face a turbulent market environment for the ten years prior to retirement (see Appendix for return and volatility assumptions). Here we see the de-risking and diversification of the outcome-focused glide path may lead to a better outcome, while the heavy allocation to equities in the final years of the Australian glide path may lead to underperformance. Similarly, the inclusion of a tail-risk hedging strategy on the U.S. market average glide path results in a tighter distribution, albeit with a modest reduction in potential returns in a more favorable market scenario. Turning to an inflationary environment, we see that the inclusion of inflation-related assets may provide for better results for the outcome-focused glide path, while the U.K.’s lack of inflation-related assets and emphasis on long-duration bonds in the final years of the glide path may reduce the projected income replacement ratio. Therefore, while risk-mitigation approaches may reduce short-term returns, we believe they are prudent and a worthy trade-off in light of a participant’s goal: to maximize the probability of meeting a real income replacement target. With this goal in mind, participants should consider trading off the probability of extremely positive outcomes to ensure the highest probability of success in all market environments.
Tail Risk Hedge Assumption: S&P 500 put options with 1 year maturity, 25% S&P 500 Index implied volatility, risk free rate of zero. Implied volatility surface is available since April, 2005. For data before 2005, a threshold factor augmented vector auto-regression model was used to interpolate the implied volatility surface based on market factors including realized volatility, daily stock return and lagged values of the model’s interpolated volatilities.
Setting a loss-capacity budget
The reader may ask “How much a plan participant could afford to lose and still retire on time and with sufficient buying power?” This number will depend on each individual’s assumed income, wage increases over time, contribution rate, years worked, retirement date and longevity. For U.S. workers, the authors asked members of the DC consultant community in the PIMCO 2014 Defined Contribution Consulting Support and Trends Survey what they believed to be workers’ 12-month loss capacity within a specific number of years prior to retirement. Figure 9 shows that the median expert among the consultants suggested no more than a 10 percent loss capacity for workers at retirement. Clearly, the risk of loss in the typical asset allocation strategies examined has the potential to go far beyond this.
In addition to capacity, we should also consider the behavioral aspects of incurring a significant loss. It may be all too easy to assume that DC investors are rational economic animals who recognize that bearing losses from time to time is an essential part of accumulating wealth for retirement. Unfortunately, the reverse often applies: Significant market downside events may trigger ill-advised changes in participants’ investment portfolios, a total cessation in contributions, or even worse, a combination of the two. Thus, assessing whether participants have the capacity to recover from losses is only one part of appraising the consequences of poorly performing financial markets. We must also consider their tolerance to stay the course of the savings journey. In the U.K., this concern has led to the creation of the National Employment Savings Trust (NEST), a new national DC plan with a low-risk foundation phase for young DC investors. In addition, the U.K. government is exploring ways to potentially ensure that participants do not suffer loss of principal on their own (employee) contributions.
In sum, DC plans in the U.S., Australia and the U.K. may benefit from better aligning asset allocation defaults to workers’ needed outcomes: purchasing power in retirement. This would likely suggest a higher allocation to real assets for both inflation hedging and risk diversification. The assessment may also suggest a need for earlier reduction in risky assets and consideration of tail-risk hedging in an effort to guard DC assets against potentially devastating market shocks.
As plan sponsors consider the appropriate asset allocation default, they also should consider the possibility that retirees may be likely to move their assets out of the DC default within several years of retirement. In the U.S., money often is cashed out of the DC plan and rolled over to an Individual Retirement Account, while in Australia members may withdraw money for travel or other expenses. Up until now, U.K. retirees generally have not had the freedom to cash out all of their DC assets; instead, they generally have had their assets shifted to government bonds and then used to purchase an annuity. However, we anticipate future U.K. annuity buying to decline materially given U.K. legislative changes to tax treatment of pension asset withdrawals. Whether these portfolio shifts are freely chosen or otherwise they underscore the need for a more outcome-focused and risk-managed asset allocation prior to retirement. This is a subject for future research.
Sara Higgins in Australia and Brent Wagner in the U.S. contributed to this paper.
Risk Factor Definitions
Broad equity (“world and equity industry” risk factor)
- World equity is a sensitivity of the portfolio to changes in the global equity markets.
- Equity industry includes exposure to 34 equity industries.
Broad Fixed income includes the following risk factors:
Duration (“interest rate” risk factor)
- Duration measures a bond’s sensitivity to a parallel shock of the par yield curve. PIMCO’s systems use a scenario- based duration calculation. Our algorithm first prices the security, and then shocks the interest rate to calculate the bond’s duration.
- Our systems generate several additional versions of interest rate duration, including proprietary duration measures such as:
– Bull duration: bond sensitivity to a fall in interest rates
– Bear duration: bond sensitivity to a rise in interest rates
– Forward secular duration: bond sensitivity to a forward-looking yield curve shift scenario specified by PIMCO’s Investment Committee
Curve duration (“slope” risk factor)
- Interest rate duration assumes a parallel shift in the yield curve. But parallel shifts rarely occur because monetary policy acts mostly on the short end of the curve, while inflationary expectations are expressed in the longer end of the curve. Therefore, the yield curve typically steepens or flattens as interest rates move.
- Our systems define curve duration as the price sensitivity of a bond to a steepening of the yield curve. Every day, each bond is priced using our proprietary pricing models and then shocked to calculate its curve duration. Our algorithm uses the 10-year bond as an anchor point and measures steepening as the change in the two-to-10 year yield spread.
Corporate or credit spread duration (“credit” risk factor)
- Credit spread duration measures the sensitivity of the bond’s price to changes in the spread of a reference single-A-rated security. Our process to calculate credit spread duration follows two steps:
– First, the algorithm calculates the sensitivity of the bond price to its own spread. This process occurs overnight and leverages our proprietary pricing models.
– Second, the algorithm translates this own-security spread duration into a duration related to the reference single-A-rated security. This mapping relies on a proprietary model that takes into account the option-adjust spread (OAS) of the bond under consideration and the OAS of the reference bond.
Currency (“high yield,” “emerging market (EM) currency” and “developed currency” risk factor)
- EM Currency includes exposure to a basket of 30 emerging market currencies.
- Developed currency includes exposure to a basket of 11 developed market currencies.
Glide path allocations