November 9, 2020
Amazingly, after 4+ years of blogging and 200 posts, I haven’t written anything about Target Date Funds (TDFs). For some folks, they are certainly a neat tool. Your fund provider automatically allocates your regular retirement contributions to a portfolio that they deem appropriate for your age and/or the number of years you’re away from your retirement date. It’s a hands-off approach for people who don’t want to think about their asset allocation and simply outsource that task to a fund manager.
But I think not all is well in the TDF world. People planning for FIRE should stay away from TDFs. But even for traditional retirees, there are some unpleasant features. Let’s take a look…
What’s right with TDFs?
First, in all fairness, I want to point out some of the features I really appreciate about target-date funds. TDFs are an easy, hands-off life-cycle asset allocation tool for people who, without any other guidance, might just “invest” their 401(k) savings in the money market option at a 0.1% annual yield. Or even worse, without the TDF option in their 401(k) plan, people might capitulate altogether and not participate at all in the retirement plan for fear of “doing something wrong.”
I also like the whole TDF philosophy because it has made it a lot more palatable for a lot of employers to “auto-enroll” employees in their 401(k) and direct those contributions. Research has shown (see here and here) that participation rates increase if new employees are enrolled by default rather than employees having to spend the effort to sign up. It’s almost hard to believe that a lack of participation is due to laziness, essentially. And more participation in companies with auto-enrollment is due to people being too lazy to opt-out. But that’s what research appears to indicate. How crazy is that? In any case, the auto-enrollment is a lot easier to sell if employers can default employees into a generally accepted and approved investment allocation. It’s a CYA measure! You lower the lawsuit risk. If Joe Shmoe finds out that his 401(k) account went down from $10,000 to $7,000, then the company simply tells him, “Hey, don’t sue us! This is what Fidelity recommended for your age group!”
Why is there a glidepath toward retirement anyway?
Glidepaths toward retirement normally involve starting with a high equity share and low bond share for young investors and shifting to a portfolio with lower risk and thus lower equity share over time as they approach retirement. See the glidepath parameters for the Vanguard TDF family with retirement dates 2015 through 2065. Notice that they still keep the 2015 fund around, so this for folks who retired 5 years ago!
If we now assume that the glidepath parameters as a function of years to retirement stay the same over time (i.e., the 2065 fund in 2065 will look like the 2020 fund in 2020), then we can plot the projected path of the asset allocation over the life-cycle of a typical employee, see below:
There are (at least) three explanations for this glidepath shape:
- People become more risk-averse as they age: If I look at my driving style 25 years ago vs. today I definitely notice that I’m a lot more cautious now. If older folks are more risk-averse they should have lower equity shares. Plain and simple!
- People want to take less risk as their assets grow: Becoming more cautious may not even be due to age alone. If you look at two investors, one with a $10,000 retirement account and one with a $1,000,000 portfolio, who do you think would be more willing to allocate 100% into equities where you’d occasionally observe a 30% or even a 50% drop? Well, measured in percentages alone both investors face the exact same risk. But a 50% drop in a $10,000 portfolio seems a lot easier to swallow than a 50% drop in a $1m portfolio. Since older investors likely have a larger net worth you can apply the “if you won the game already, stop gambling” logic and push for a slowly declining equity weight as investors age. Thus, a lower equity share might be justifiable even if young and old folks don’t differ substantially in their risk parameters.
- A changing “implicit” asset allocation over the life-cycle: Related, but not necessarily identical to the points above, one of the reasons why younger investors should take more risks with their portfolio (i.e., a higher equity allocation) is that they have more “runway” ahead of them in the form of future cumulative earnings potential. A stock market drop is much easier to swallow if you have 30 more years of future contributions than for an investor who is 3 years away from retirement. In other words, one could view the cumulative sum of future contributions as an “implicit bond portfolio.” And the equity glidepath in the financial portfolio is there to counterbalance the shrinking implicit bond portfolio as the investor ages.
So far, so good. Let’s take a look at eight reasons why I still dislike TDFs:
Dislike #1: An additional layer of fees
Most of us in the FIRE community are extremely cost-conscious. Almost maniacally cost-conscious. We don’t like actively-managed funds with high expense ratios and prefer low-cost passive index funds instead. So, if you ever even consider a target-date fund make sure that the underlying components of it are all passive funds. That would be the case with the Vanguard TDFs. Fidelity apparently has two TDF fund families one with expensive managed funds and one with passive funds only. For example, for the 2030 TDF, they have the FXIFX with a 0.12% annual expense ratio, but also FFFEX with a 0.68% expense ratio.
But even with Vanguard, the TDFs have an additional layer of fees. For example, the Vanguard expense ratios for the 2015 to 2065 TDFs range from 0.13% to 0.15%. If you had built the same asset allocation with the underlying funds yourself you’d pay only 0.07%. You could even move your money to Fidelity and achieve the same result with an even lower expense ratio. All the way down to 0.003% for the 2045 to 2065 funds, thanks to the zero-expense-ratio funds for both domestic and international equities!
Of course, 0.10% to 0.15% may not seem like a lot of money, but most (all?) of us in the FIRE community strive to accumulate seven-figure retirement accounts, and in a million-dollar account, that would be $1,000-$1,500 a year. Why would you want to pay that kind of money every year to someone just shifting around a few fund allocation weights every once in a while? And again, this is not active management, not stock picking, not market timing, nothing that requires an army of finance, economics, and physics PhDs to run. It’s merely rebalancing a portfolio of about 5 mutual funds back to the target weights. Every DIY investor, every Bogelhead, every FIRE disciple should be able to do this!
Dislike #2: Potential Tax Inefficiencies
Most of the time, TDFs are held in tax-advantaged accounts. In fact, as I stated above, you’ll most often find them as the default allocation in your 401(k) plan. If you hold your TDF in your 401(k), there’s no tax inefficiency. Not so in a taxable account, though. Because of the frequent rebalancing in the TDFs due to 1) rebalancing back to the target weights and 2) changing the target weights over time, you’ll generate taxable events. Not only that, most of the taxable events will be capital gains:
- Rebalancing back to the target weights means that you sell winners and buy losers.
- Lowering the equity weight over time means that you will likely sell equities who have a higher potential for generating capital gains than bonds.
Dislike #3: How do you handle your non-retirement accounts?
For those of you who objected that item #2 is no big deal because it only impacts taxable accounts and you will hold your TDFs in a tax-free or at least tax-deferred retirement account, there’s still a headache I like to point out to you. Let’s assume for a second that you’ve convinced yourself that the Vanguard TDF asset allocation path is indeed your favorite and accepted asset allocation over the life-cycle. I don’t think it should be (more on that below) but let’s ignore the pesky objections from cranky old Big Ern for now. Most of us, especially in the FIRE community, will also have regular taxable non-retirement brokerage accounts. If our retirement account is invested 100% in your TDF, then by simple arithmetic, your taxable account has to match the TDF allocation as well in order for your overall allocation to match your target allocation. So, you do introduce the tax efficiency headaches into your taxable account after all.
You will also likely run afoul of some of the tax arbitrage analysis I performed in Part 35 of the Safe Withdrawal Series. There, I made the point that it’s highly unlikely that you want to keep the same asset allocation in both taxable and retirement accounts. Let’s look at the following example. Imagine an investor who has $100,000 in a taxable account and $100,000 in a retirement account and wants to implement an 80/20 overall allocation at the current spot on your glidepath. If bond yields are high and you face a high marginal tax rate on ordinary income (bond interest = ordinary taxable income) you will likely want to hold the entire bond portion of $40,000 in the retirement account. So you would end up with a 100/0 portfolio in the taxable account and a 60/40 in the retirement account. The classical recommendation of holding tax-inefficient assets in the retirement account. (side note: we’d also have to factor in the different after-tax values of $1 invested in taxable vs. tax-deferred vs. tax-free accounts, but let’s abstract from that for this simple example!)
Conversely, if bond yields are low and/or you face a low marginal tax rate you might benefit from the White Coat Investor recommendation of holding the low-yielding asset in the taxable account, which means you have a 60/40 allocation in the taxable account and 100/0 in your retirement account.
Another headache would be the REITs allocation in some of the TDF families. REITs are extremely tax-inefficient because dividends are considered non-qualifying on your U.S. tax return and thus taxed at a higher rate. So, if you prefer a TDF allocation with REITs in your overall allocation, you’d be smart to keep all of your REITS in the retirement accounts and reshuffle the rest to fit the overall asset allocation.
Consequently, even if you really prefer the TDF asset allocation on an aggregate level you’ll likely leave money on the table through tax inefficiencies unless you get your hands dirty and maintain and rebalance the whole shebang yourself not just within but also across different accounts and especially account types; taxable vs. tax-deferred vs. tax-free. So for example, if your overall allocation drifted too far away from the target because equities performed strongly, then you don’t liquidate equities in your taxable account, but rather do the rebalance entirely in the tax-deferred account. So you’d have to overcompensate in the retirement account because you’re stuck with the overweight in the taxable account. Which defeats the whole purpose of simplicity in TDFs, doesn’t it?
Dislike #4: Young investors should have 100% equities
I’ve looked at various different TDF families from different providers and one feature sticks out: They never start at 100% equities. The highest equity shares are just about 90% at the large TDF providers: Vanguard, Fidelity, and T. Rowe Price. Even lower in some other fund families. I would normally recommend that investors who are just starting out on their path to retirement savings should just keep it simple and put 100% into equity index funds. Nobody can tell me that a 25-year old making $50,000 can’t take a 100% equity allocation in their $10,000 retirement account. This person has another 40 years of work-life and contributions, another 5 or so recession/expansion economic cycles, and probably more than 5 bull/bear market cycles ahead of him/herself. My personal opinion is that a 10% bond allocation is 99% fig leaf and CYA on behalf of the plan administrator and 1% diversification benefit.
For full disclosure: Some experts find that “Target-Date Funds Are Too Risky for Savers” (Yahoo! Finance) but that’s not based on any actual analysis, just on “feelings”. To make this less about feelings and more about analysis, let’s take a look at how “Optimal” glidepaths would look like pre-retirement. I put the “Optimal” in quotation marks because nothing is really optimal, it’s all subject to the objective function and the assumptions you make along the way and subject to another really big bummer, see item #7 below! In the chart below, I contrast the Vanguard TDF to my own optimization calculations. I don’t want to get too much into the weeds of my simulations and optimizing, but for the math wonks:
- It’s a Monte-Carlo simulation
- There is a concave (risk-averse) utility function over the final payoff. Constant Relative Risk Aversion (CRRA) with a gamma parameter at 3.0.
- I don’t search over the entire universe of possible glidepath shapes (a gazillion different combinations!) but restrict myself to a start point, an endpoint, and several “kink points” along the way so I’m able to simulate flat spots and concave or convex shapes.
So, how do “optimal” glidepaths look like? My initial “Optimal” GP kept equities at 100% until 20 years before retirement and then shifts down to 60%. But notice that the entire red line is always above the Vanguard GP. Maybe I didn’t set the risk aversion parameter high enough and that explains the 100% initial allocation. So, then I jacked up the risk-aversion to a pretty unrealistically high level and get the yellow line. And I’d indeed push the final equity share to 40%, well below the Vanguard 50%. In fact, this very risk-averse investor would spend about 27 years before retirement with a lower equity share than under the Vanguard assumptions. But the initial equity allocation still stays at 100%! Even the craziest risk-averse investors would still want to start with 100%. And the rationale, again, are the two points I made in the intro. 100% equities might imply a high risk in percentage terms. But in dollar terms, it’s still relatively low-risk, considering the big chunk of future contributions, which almost works like an implicit non-shortable bond allocation when starting out. (and yes, yes, yes, I’ve done a lot of sensitivity analysis and the 100% starting point is indeed quite robust. I am not going to get into the details here but it’s something I worked on 4 years ago while still advising clients on this issue)
So, my suspicion is that a lot of the industry TDF shapes are based on two assumptions baked into their “optimization”:
- A more than average risk-averse investor,
- An external constraint on the initial equity weight at about 90% (all the way down to 80% for some other TDF providers) that’s based not on hard science but mostly on legal and behavioral considerations.
Neither of the two applied to me. Nor should they apply to the FIRE-community. TDFs are calibrated for and, pardon the pun, “targeted” at a different audience!
Dislike #5: TDFs reduce the equity allocation too early
Another pet peeve I have with TDFs is that the equity share starts moving down too early. That’s the case both in the actual industry TDFs and also my Monte-Carlo simulations and optimization exercises. Imagine you just found the FIRE community and you want to employ a TDF on your path to early retirement in 10 years. Your Vanguard 2030 fund would currently allocate only 67% to equities and the remainder to bonds with absolutely pathetic yields currently below the inflation rate. And then your TDF would slowly shift to only about 50% stocks along the way.
I don’t think this makes a lot of sense for us in the FIRE community. I agree that you might want to hedge the risk of a market downturn if you are a traditional retiree at age 50 or 55 with a sizable nest egg and a savings rate of “only” 10-15% where you have less Dollar-Cost-Averaging opportunities to make up for major losses. But if you start out with $0 and a 50-60% savings rate you should have a much more aggressive equity share!
One workaround for this problem would be for FIRE savers to pick a TDF with a retirement date probably around 10-15 years after your planned retirement. For example, it’s 2020 now. A new FIRE enthusiast who likes to retire in the year 2030 would pick a Vanguard TDF with a retirement date of 2040 or 2045. Now you’d start out with 80-90% equities and would end up at around 67-75% equity share at your retirement date. The 67% to 75% equity share at the retirement date is not that far off from what I’d normally recommend for retirees both in the FIRE and traditional retirement scene. Yeah, the initial equity allocation is “only” 80-90%. I’d prefer 100% but if you must use the TDF, then so be it.
Dislike #6: TDFs are potentially a bad strategy for people with a late start
Even for traditional retirees, TDFs may be overly simplistic. What if you get a late start to retirement savings? Imagine you’re a 50-year-old starting with zero retirement savings. It could be because you’ve neglected your retirement savings so far or you had some other life-changing event (business failure, divorce, medical bills, etc.) that reset your net worth and makes you start from scratch again. You face the same problem as the FIRE investor above. Starting with such a meek asset allocation will never get you to a comfortable retirement. Drastic circumstances call for drastic measures; you might have to aim for a higher equity share. Remember, the 75% equity share for the 2035 TDF is calibrated to a retirement saver who already has a sizable nest egg. It may make sense for that retiree to take some chips off the table but not for someone just starting out!
Dislike #7: TDFs violate an important mathematical principle!
For those of you with an economics or engineering or math background, you’ll agree that some of my objections to TDFs above can be succinctly summarized into:
Target Date Funds violate Bellman’s Principle of Optimality
That principle is, let me quote:
“An optimal policy has the property that whatever the initial state and initial decision are, the remaining decisions must constitute an optimal policy with regard to the state resulting from the first decision.” Bellman, R.E. (1957), Dynamic Programming, Dover.
In other words, if you optimize the GP the way it’s currently done, setting equity/bond allocations merely as a function of the number of years to retirement, you will violate Bellman’s Principle. Take two investors, both 45 years before retirement, starting out with $0 initial assets and contributing $1,000 each month to their retirement plan TDF. They have an identical objective function over the final wealth. 10 years into their path, investor A has $100,000 and investor B has $300,000 in savings (because he/she started at a different time with better returns or got an inheritance). If we now re-optimize another glidepath for each investor with a 35-year horizon (again, the GP depending only on age), the same final value objective function but different initial net worth numbers ($100k vs. $300k) you can show that those glidepaths will have to differ because of the different initial wealth. At a minimum, one investor, probably both investors will now have reoptimized glidepaths looking different from the original glidepath for years 11 through 45. A direct violation of the Bellman Principle.
The mathematically correct way to construct a TDF would be to allow for a lot more “path-dependency” of glidepaths along the way, i.e., your asset allocation depends not just on your age but also on the asset level. This would be a stochastic dynamic optimization problem, which is actually not that complicated to solve. You would optimize this by “backward induction,” something that’s done routinely in economics and finance. But it might be a bridge too far for the typical 401(k) plan administrator! And since the lawyers will not understand it, we are stuck with the sub-optimal, one-size-fits-all TDFs.
Dislike #8: TDFs use the wrong post-retirement glidepath
At least I agree with the pre-retirement glidepath shape qualitatively, though not exactly quantitatively. The equity glidepath moves down as you age and we can argue and haggle how steep the slope should be. It’s a matter of a quantitative disagreement.
But that changes after retirement. Most TDFs shift further out of equities as evidenced in the Vanguard TDFs (49% equities at retirement start, 34% when 5 years into retirement). The same is true for most other fund families I’ve checked as well.
But Michael Kitces pointed out that around your retirement date you should ideally have a “bond tent“: shift more into bonds before retirement and then shift out of bonds again in retirement. This is a viable solution to alleviating Sequence Risk. Not a perfect hedge – nothing ever is – but this glidepath out of bonds and back into equities certainly hedges a little bit of Sequence Risk. I confirmed that result in my simulations in Part 19 and Part 20 of the Safe Withdrawal Series. I got slightly different quantitative results because I rely on historical returns and Kitces/Pfau on Monte-Carlo simulations. But the same flavor prevails: early in retirement, you want to keep a lot of bonds around as a hedge against a bad bear market. But then shift back into stocks to make sure your assets last over the next 30 to 50 or even 60 years.
This one problem in the TDF design may not be the worst dealbreaker because there is an easy workaround. You most likely hold your TDF in a retirement account so if you’re happy with its allocation up until retirement but not during retirement, simply sell the the TDF at the end of your career and switch over to a mix of equity and bond index funds more in line with the Kitces/Pfau and ERN glidepath model. Or at least keep the allocation constant. And if your 401(k) plan doesn’t offer the inexpensive index funds, simply roll your account over to an account at Fidelity where you do have access to zero-expense-ratio funds.
While some form of a glidepath shifting down the equity weight on the path to retirement may be desirable I don’t really care much for actual TDFs available today. Certainly, folks in the FIRE community should stay away. But there are even some headaches for traditional retirees.
Also, this is not the last time I want to write about this topic. I’m particularly interested in the pre-retirement glidepath under different assumptions: Monte-Carlo simulations vs. historical returns. The impact of the initial portfolio value on the “optimal” glidepath. Different objective functions. How do glidepaths that do satisfy the Bellman Principle differ from the sub-optimal conventional glidepaths? I hope to address all that in one or even multiple future blog posts. If I find time in my busy retirement schedule to finally put that together! 🙂 Stay tuned!
Hope you enjoyed today’s post! Please leave your comments and suggestions below!