I am on the record in opposing the addition of a real estate index fund to the Thrift Savings Plan’s existing five-fund lineup. While I have nothing against real estate investment trusts (REITs) as investment vehicles and I am generally in favor of an investor’s right to choose, in this case more is not better.
There are costs and risks associated with such a significant change in TSP options, not the least of which are the creation of a precedent for using TSP as political capital and compromising the interests of plan participants.
The most important factor to consider in judging the value of a new fund option is the benefit it is expected to deliver to plan participants. To evaluate the real estate fund option, I’ve fired up the Monte Carlo simulation engine — a type of scenario analysis that repeatedly simulates many individual lifetimes, usually thousands, while varying the annual investment returns during each simulation. In this case, I compare some expected results for two hypothetical employees.
One is a younger Federal Employees Retirement System participant with 20 years until retirement, and the other is an older participant who is retired and taking TSP withdrawals on a regular basis. These hypothetical cases have been simplified in order to highlight the effects of adding a REIT allocation on the expected withdrawal rate that can be supported from two different portfolio allocations, one aggressive and one conservative.
In both cases, I based assumptions about the expected returns from the various TSP funds and a new REIT index fund on historical averages dating back to 1926 for the asset classes they represent. These expected rates of return were then reduced by 0.1 percent to account for the funds’ expense burden.
Specifically, the expected annual return for each fund is about 12.3 percent for the C Fund, 16.2 percent for the S Fund, 13.3 percent for the I Fund, 5.9 percent for the F Fund, 6.0 percent for the G Fund and 9.8 percent for a REIT fun.
For the REIT fund, I chose to use the aggregate data for a broad range, rather than a particular type, of domestic real estate, since this produces a more conservative estimate.
These expected returns are then combined, in the appropriate proportions, to produce the expected rate of return for the asset allocation to be considered. Correlation and risk factors are also considered in the estimates.
Case 1: The participant is 40 years old, earning $50,000 per year, has $100,000 in his TSP account and is contributing 10 percent of his salary to his TSP account. His agency is contributing another 5 percent on his behalf. He will work and save to age 60 and then spend 30 years in retirement. His goal is to maximize his regular annual withdrawal rate from his TSP account during retirement while keeping his risk of running out of money during the withdrawal period low. He invests his account according to TSP’s most aggressive L Fund allocation — 42 percent in the C Fund; 25 percent, I Fund; 18 percent, S Fund; 10 percent, F Fund; and 5 percent G Fund. If he maintains that allocation for life, history indicates he should expect to earn about 12.1 percent per year. He can expect to withdraw $30,000 per year, adjusted for inflation, and die with $3.3 million, in today’s dollars, in his account. This is the most likely ending
balance, although the range of possible outcomes varies widely.
The most aggressive plan proposed by supporters of adding a REIT fund to the TSP options is to allocate 20 percent of contributions to a REIT fund. Doing so and proportionately reducing the allocation of each of the other five TSP funds reduces the expected volatility of the portfolio, but also decreases the rate of return it will most likely earn to 11.7 percent. The result is that the investor’s maximum annual withdrawal falls to $29,000 per year and the most likely ending balance for the account falls to $2.7 million. The probability of failure remains low and is identical to the non-REIT scenario. So, in this example, adding a REIT fund to the investment allocation reduces the participant’s expected withdrawal rate and the expected ending account value — a negative outcome. The effects of adding REIT allocations smaller than 20 percent are similar and proportional to the size of the allocation.
Case 2: The participant is 80 years old, retired, has $280,000 in her TSP account and will spend another 10 years in retirement. As in Case 1, her goal is to maximize her regular annual withdrawal rate from her TSP account during retirement while keeping her risk of running out of money during the withdrawal period low. She invests her account according to the most conservative L Fund allocation — 12 percent in the C Fund; 5 percent, I Fund; 3 percent, S Fund; 6 percent, F Fund; and 74 percent, G Fund. If she maintains that allocation there for life, she should expect to earn about 6.1 percent per year, and can expect to withdraw $30,000 per year, adjusted for inflation, and die with $38,000, in today’s dollars, in her account.
In this case, adding a 20 percent allocation to a REIT index fund and proportionately reducing the allocation of each of the other five funds increases the expected volatility of the portfolio and increases the rate of return it will most likely earn about 6.9 percent. This is because the REIT fund’s volatility and expected rate of return are higher than those for the F and G funds, which together make up 80 percent of the base allocation. The result is that her maximum annual withdrawal rises to $30,500 per year and the most likely ending balance for the account rises to $48,000. Once again, the probability of failure remains low and is identical to the non-REIT scenario. So, in this example adding a REIT fund to the investment allocation increases the participant’s expected withdrawal rate and the expected ending account value — a positive outcome. As in Case 1, the effects of adding smaller REIT allocations are similar, though less pronounced.
There are costs and risks associated with adding the REIT fund, though, so it should logically be avoided unless the benefits outweigh the costs and appear to be worth the risks. For younger, more aggressive investors, there is no meaningful benefit to using a REIT fund as part of their portfolios, and no compelling reason to add one to the TSP’s investment options. A REIT fund does produce additional benefits for an older, more conservative participant, however, and this participant might well support the addition of a new fund to the plan.
But, there is another option that avoids the negative implications of adding a new fund — reallocating the existing funds to produce the same result as the portfolio containing the REIT allocation. Instead of using a 20 percent REIT allocation, suppose the participant in Case 2 reallocated the existing TSP funds in a way that produced similar results — an expected annual return of 7.0 percent.
A little trial and error reveals that by allocating 10 percent of her account to the C Fund, 10 percent to the S Fund, 5 percent to the I Fund, 25 percent to the F Fund and 50 percent to the G Fund and maintaining this allocation for life, she could expect to withdraw the same $30,500 each year and end with the same $48,000 balance.
So, based on what many participants may consider the two most important benefits of investment performance — the ability to maximize expected rates of withdrawal during retirement and the terminal account balance — it is possible to achieve similar results with or without adding a REIT to the TSP.
Written by Mike Miles
For the Federal Times
Publication May 1, 2006