Friday, December 8, 2017

The Return You Need

A topic that frequently appears in the comments section of my posts compares market returns to annuity or Social Security retirement benefits (also an annuity). The comment is usually some form of “I would only need a return of x% on my investment portfolio to outperform purchasing an annuity (or some other low-risk alternative) today.”

One of the most common mistakes I see retirees make is looking at a good return on a nearly risk-free investment and concluding that they would do better in the stock market because it has a higher long-term average return.

This is a good opportunity to review liability matching.

Long-term market return averages are interesting when we have an investment portfolio that isn't used to fund specific future liabilities (expenses).

Once we begin to spend periodically from that portfolio, we introduce sequence-of-returns risk[2]. We're trying to fund many expected future annual expenses and not knowing the price we will receive for those future stock sales introduces risk. The sequence of returns is often more important than long-term market return averages.

On the other hand, when we are saving and investing in stocks for a single future expense, like buying an annuity or paying for a wedding, we can't be sure that we will be able to sell the stocks at the price needed to pay that liability when it comes due. We may know how much the wedding will cost and when it will be held but we won't know the future value of our stock investments until the big day arrives. This doesn't introduce much sequence risk because we only sell one time instead of every year. This a "duration-matching", or "liability-matching", problem.

Two recent comments effectively asked what market return would be needed to make buying an immediate annuity (SPIA) in the future a better deal than buying a deferred income annuity (DIA) today. One commenter wants to claim Social Security benefits early, invest them in stocks, and use the proceeds to purchase a SPIA in four years. He asks what average market return would make this strategy come out ahead of postponing claiming benefits, which would increase them by 8% a year (simple interest).

The second reader wants to postpone buying a DIA, invest the purchase money in stocks, and buy a SPIA after 15 years. He believes that if he can earn 5.37% or so on stocks, he will break even on payouts.

Both questions are posed as the rate of stock returns these strategies will "need" to provide outcomes identical to buying a DIA now or postponing Social Security benefits. The needed rates of return, however, aren't the key to this analysis and besides,  the fact that you "need" a certain return, in my experience, doesn't make you more likely to earn it. Mick told us that if we try sometimes we might find we get what we need. Notice the words "sometimes" and "might."

Before we consider liability-matching, let me point out a few other concerns.

First, the presumed ultimate goal of these strategies is to purchase future "safe" income in the form of an immediate annuity or larger benefits. Does it make sense then, to expose the assets you have earmarked for safe income to stock market risk for four, 15 or any other number of years first?

It's a little like saying, "I really need to buy some very dependable income with this money but I think I'll bet it at the racetrack first because if I win I'll be able to buy even more safe income!" You need to consider other possibly less attractive outcomes.

Second, DIAs are significantly cheaper than SPIAs[3] and postponing Social Security benefits is an even better deal than DIAs. So, that means you not only have to outperform the 8% additional Social Security benefit for every year you postpone to age 70, but you also have to offset the cost advantage of DIAs or postponing Social Security benefits.

Third, buying a SPIA today or postponing benefits are relatively risk-free strategies — you know exactly how much income you will receive and when — while a stock investment is quite risky. You don't really "break even" when you earn the same benefit through a much riskier strategy.

If you wade through a swamp to get a cold beer, your buddy takes a boat, and you both end up with identical bottles of beer on the other side, your buddy got a better deal. Way better.

Finally, Social Security benefits are adjusted for inflation and most annuities aren't. You can purchase inflation-protected annuities but they are quite expensive and that makes the SPIA even less competitive in this analysis. It's very difficult to match the price or features of Social Security benefits on the annuities market.

A great example of what could go wrong with these strategies is available as recently as the Great Recession from late 2007 to early 2009. Stock indexes fell more than 50%.

Imagine that you employed the four-year early-claiming strategy in 2005 and ended up with half the money you needed to buy that SPIA in 2009. Or, imagine you implemented the 15-year strategy in 1994. The long-term market return average would have been irrelevant. The problem would have been needing to pay the bill at a market bottom.

Which brings us back around to liability matching, which offers the solution to this problem by matching the timing of a liability to the duration of the investment that will fund it.

I've talked about duration before, but here's a simplistic explanation. Duration measures the number of years that an investment needs to recover from a loss. The longer the asset's duration, the longer it takes to recover from losses.

If I know I will need $1,000 for an expense in four years, I can buy a $1,000 Treasury bond that matures in four years and know with relative certainty that I will have a thousand dollars to pay that expense in exactly four years. How much money will I have in four years if I invest in stocks, instead? That's unknowable. And, stocks have durations often measured in decades.

What if I buy bonds that mature in two years, instead of four? Then I will earn less interest because shorter bonds pay less. Ideally, the bond's maturity (equalling its duration as it nears maturity) will match the timing of the expense. Too short and we earn less, too long and we can't be sure of its value when the expense arrives.

The proper investment vehicle to fund this 4-year strategy would be short-term bonds that currently earn a percent or so, not stocks. Clearly, a percent or so return won't outperform postponing benefits. Investing in stocks might, but is it a high-risk strategy. It will sometimes miss badly.

"Might." "Sometimes."

Likewise, the 15-year investment to avoid purchasing a DIA today should match liabilities. That means investing in stocks for at most the first five years and then in much lower return bonds for the final ten years. Even if you're willing to risk your "safe money assets" in the market for five years before buying safe income, the average return you would need on stocks, with ten subsequent years of today's low bond yields, would be nearly 13%, not 5.37%.

And still, you couldn't know how much your stocks would be worth at the end of that five years, or your total capital after 15.

So, to summarize, investing in stocks to meet a known future liability like purchasing an annuity or even retirement itself is a risky strategy. It may be fine to invest in stocks when the liability is a decade or more in the future, or when there is no specific liability. But, as the liability's due date approaches, its far safer to begin shifting to lower duration, lower yielding, liability-matching assets.

Just ask the many workers who had to postpone retirement in 2009 because they no longer had the amount of savings they needed. That's why we recommend cutting your stock exposure the decade before retirement. They had a date when they expected to need retirement savings but they had to postpone retiring for several years until the market recovered because they were holding too much stock.

And, don't forget Occam's Razor[4]. In my experience, cute, complex or tricky strategies to fund retirement are generally flawed. The simplest answer is generally the best. Retirement advice contains many strategies that are "too cute by half."

There is a role for stocks in an adequately-funded retirement plan; providing secure income isn't it.

Take the boat.

Long-term average stock returns aren't the key issue.  The problem is market bottoms messing up your wedding. (And you were worried about rain!)

The analysis isn't as simple as comparing expected returns. You can drown in a river that averages a foot deep and you can go broke in a market that averages 9% returns over the long-term.


[1] The Retirement Café: Clarifying Sequence of Returns Risk (Part 1)

[2] What is Duration? Investopedia

[3] Financial planner, Tom Morris, provided some quotes for a $1M DIA at age 62 today and a $1M SPIA at age 70. The DIA, if purchased today, would pay out $5,551 monthly when the 62-year old reached age 70. The SPIA would immediately pay out $3,227 monthly to a 70-year old if purchased today. The probability that a male aged 62 will live to receive income from the DIA is 89%. The probability that at least one spouse of a couple would is 99%.

[4] Occam's razor.


Some older posts at The Retirement Cafe didn't convert well when I changed the format a few years back. I'm told some are difficult to read. I may someday go through hundreds of posts to find and correct them (really, it could happen).

In the meantime, they can easily be read from your mobile device which displays a black-on-white rendering. Or, if you prefer your computer, click the "View the mobile version" link at the top right of this page.

Alternatively, if you enter your email address in the Follow by Email box you will receive a black on white email version each time I post in the future, though that won't fix the older posts.

Thanks! And, sorry for the inconvenience.


  1. Just out this week.

    1. Thanks for sharing this, Jack. Given the target, middle-income retirees who haven't saved a ton of money for retirement and who don't want to consult a planner, this seems like very good advice. It is a simple and safe approach.

      One can do better than RMD for a spending strategy, but for an underfunded retiree who doesn't seek out an advisor, I think it would be hard to improve upon.

  2. I'm considering an SPIA primarily because the bond portion of my wife and my nest egg seems about as productive as putting it under the mattress.

    Let's assume we proceed. We're within a few months of each other in age (65) and both in good health with long-lived ancestors.

    Is it more efficient to purchase one SPIA for both lives or to use half the funds to purchase for each of us a separate SPIA? That second alternative would seem to be identical to a joint SPIA with the continuing payment at 50% upon the first death.

    I'm hung up on the poor IRR my HP 12-c shows for the joint version. When I ran it last year, the IRR didn't reach 3% until payments had been made for 30 years. I can get a CD promotional rate of 2.5% already and I can't imagine it going anywhere but gradually up over the next 30 years.

    1. Ron, the purpose of your bond allocation is to dampen volatility, not to increase your portfolio return, i.e., not to "be productive". From late 2007 to 2009, stocks fell 51%. A bond return of even 0% would have outperformed stocks by 51%. Figure out how much you want to allocate to safe investments and accept the fact that you will receive a minimal return on that portion of your portfolio but that it will soften your fall in a bad market.

      I believe the general recommendation for a couple near the same age is for a single joint-and-survivor annuity, but there may be other considerations. Some annuities lower payouts at the death of the first spouse, as your proposed arrangement would, but traditional life annuities lower the payout only when the primary annuitant dies. It depends on which you buy.

      IRR isn't an effective way to evaluate a life annuity because an annuity's IRR depends on how long you will live. Here's an explanation from Wade Pfau:

      To know the annuity return, it is necessary to know how long the annuitant will live and receive payments. Or, at least, returns can only be calculated by assuming how long income payments will be received. I generally do not speak about the returns from annuities, because for a life-contingent annuity you will not know your overall return until you know how long you have lived. The longer you live, the higher the return from the annuity. For life-only annuities, returns will be very low early on as relatively little income has been received relative to the principal purchase amount. Eventually, the return will exceed the payout rate.

      Lastly, I find that a good insurance agent with retirement planning expertise is worth his or her weight in gold when evaluating annuities. The devil is in the details and annuities have a lot of details. I generally refer clients to one and they have all seemed quite happy with the results.

      Thanks for writing!