A recent survey conducted by our organization of leading plan sponsors, consultants and target date fund sponsors found significant support for the further addition of “alternative” strategies in defined contribution plans, including both listed and non-listed real estate.
The survey, which consisted of in-depth interviews with industry professionals collectively responsible for about $14 trillion in assets, left open the question of how best to add real estate assets to these plans, and the practical impact of increasing exposure to this asset class on plan participants.
Though many plans already include some exposure to public-traded real estate in the form of Real Estate Investment Trusts (REITS), including non-listed real estate has proven to be a more difficult challenge. The issues are well known – primarily concerns about valuation and daily liquidity and, in some instances, fees.
Nonetheless, direct real estate has a number of attributes that should make it attractive to DC plan sponsors. Two in particular stand out: the ability to add diversification, and returns that generally demonstrate a low correlation to the broad equity market. As publicly traded securities, REITS often act more like equities than real estate. Non-listed real estate, on the other hand, demonstrates a low correlation with equities.
Why a low correlation is important
New research sponsored by our group shows why this low correlation is important. The report’s authors looked at the impact of adding a 10% allocation of listed and unlisted real estate to a range of DC plan retirement outcomes.
The study used a data sample covering the period from January 1976 to January 2014 and a range of indices as a proxy for the traditional assets classes (MSCI All Country World – excluding the US, the Barclay’s US Aggregate and Barclay’s Global Aggregate, and the FTSE NAREIT U.S. Real Estate Series, for example). A block bootstrap approach was used to simulate the data.
For purposes of the simulation, the researchers looked at two hypothetical plan participants; a 25-year old worker with a starting salary of $40,000 and zero in retirement savings, and a 40-year old worker with a salary of $53,835 and $100,000 in retirement savings. Each worker then experiences salary increases of 2% per year, which synchronizes the salary level of each worker (the 25-year old worker’s salary will be $53,835 when they turn 40-years of age). Both workers contribute 8% per year of their salary to a retirement plan portfolio on a continual basis throughout each working year. The income of each worker in his or her final year of work prior to retirement was $88,322.
The authors conducted the simulation for each worker using terminal portfolio values and the retirement wealth ratio (RWR) as a means for comparison, with RWR the ratio of terminal wealth to final salary. RWR was selected as the preferred metric to define success because it is a tangible outcome-oriented measure, translatable into dollars, and relevant to all the audiences addressed by the study. Broadly, success was defined as a RWR equivalent to 12 times final salary.
To gauge the impact of adding real estate to the portfolios, they looked at simulated designs with asset allocations that ranged from 100% stocks and 100% real estate to 60/40 stock/bond blends.
Several popular target date and target risk funds were also included in the simulation. The 100% stock portfolio was included to provide a benchmark for a long-term wealth maximizing strategy; the 100% real estate portfolios were incorporated to determine what the end result would be for someone with only exposure to this asset class.
The 100% Real Estate Blend portfolio demonstrated returns that were less than those of stocks, but better performance was seen in the worst scenarios. For example, that the 100% Real Estate Blend portfolio provided the highest conditional value at risk (CVaR) for the 25 year-old plan member (i.e. the individual with an initial 40-year horizon) suggesting that the very worst outcomes for this strategy are superior to those of the others analyzed. However, the chances of such a portfolio being used as the only investment in a DC plan are very remote.
They then considered the performance of two target date, or lifecycle, funds using two well-known funds as proxies. One fund maintained an initial allocation of 90% stocks for the first 15 years of the glidepath, with a "landing point" at retirement of 55%; the second fund began with similar stock exposure but ended at 40% stocks. The researchers then looked at what would happen if they added the real estate blend investment to each glidepath. In the examined scenario, the allocation to the real estate blend remained constant at 10% throughout the time horizon, whereas stock weightings fall and bond weightings rise.
As observed with the balanced strategies, there was an improvement in downside performance. In contrast to the target risk strategies, median outcomes improved slightly too. The researchers concluded that as the portfolio size effect became manifest – that is, in the last half of the accumulation phase before retirement – the superior performance of the real estate blend strategy over bonds compounded, thus, benefitting returns.
The observed benefits of adding the real estate blend appeared to occur regardless of the age of the participant. For example, the results for the 40-year-old investor appeared to mirror the results of the 25-year-old, indicating that similar expected performance is possible, but with better downside performance.
The most noticeable impact was the difference between the absolute value of each measure for the same strategy. For the 100% stocks strategy, median wealth is 64% higher over the 40-year time horizon ($6,136,084) compared to the 25-year time horizon ($3,742,004). In contrast, for the 100% bond strategy median wealth is 0.4% higher over the 40-year horizon ($847,152) compared to the 25-year horizon ($843,426). As expected, the blended portfolios fall on a continuum between these extremes, with the differences again accounted for primarily by the effect of compounding on a rapidly growing portfolio.
Don’t discount the behavioral impact
Compounding is a function of both time and investor persistence. Time, of course, passes on its own but maintaining a consistent exposure to the market is a function of investor discipline. It’s with this in mind that the authors highlight the importance the timing of portfolio returns plays in the ability of DC plan participants to achieve their goals, particularly in the later phases of accumulation and the transition to retirement. It follows from this that reducing portfolio volatility is critical to addressing the all-important behavioral side of the equation, helping investors stay the course and providing a smoothing path to retirement.
Plans sponsors have acknowledged an interest in expanding exposure to alternative asset classes, including real estate, primarily as a means of achieving higher levels of portfolio diversification. The 10% real estate allocation utilized by the researchers appears to serve exactly this purpose for the vast majority of portfolios likely to be found in a DC plan. The reported concluded that with the real estate allocation participants achieved similar expected outcomes to those of comparable portfolios without real estate, did so with better tail risk characteristics, and, realized success to a similar extent as their non-real estate alternative portfolios, but with a smoother path to the end goal.
Laurie M. Tillinghast and David Skinner are co-directors of the executive committee of the Defined Contribution Real Estate Council.