Get Rich Slowly recently hosted a post from Motley Fool writer Robert Brokamp about how much money it takes to retire. The article links to the Motley Fool’s “Am I Saving Enough?” calculator, which seeks to answer the question of how long you can expect your retirement savings to last.

I’ve written before about why I don’t trust retirement planning calculators, and this one is a perfect example. Go ahead and take a look at it.

Among other things, it asks you:

- How much you have saved now and how much you’re saving per month,
- What portion of your savings are in which type of accounts (401k, Roth IRA, taxable, etc.),
- How much you expect to spend each year in retirement (and it allows you to provide a good deal of detail for how you expect that figure to change over time),
- Your current tax bracket as well as your projected retirement tax bracket,
- How much you expect to receive from social security or a pension, at what age you expect to begin receiving such income, and whether or not that income is adjusted for inflation,
- How old you are now, at what age you expect to retire, and what age you expect to live to.

You get the idea. It asks for much more information than most retirement calculators. This is a good thing, as it allows for greater precision.

### But It’s Still Worthless.

Unfortunately, the answer the calculator gives you is complete garbage.

The reason — and I bet you saw this coming — is that the calculator uses unrealistic assumptions for its calculations. Specifically, it asks you to enter a given rate of return, and it then assumes that your portfolio earns *that same return every single year*.

### Let’s Be Realistic.

In real life, returns vary from year to year — even if you stick with extremely low-risk investments. A calculator that assumes a constant return every single year is going to significantly understate the amount you need saved before you can retire safely.

Said differently, if you expect your portfolio to *average* a certain rate of return over the course of your retirement, you probably need to set your starting withdrawal rate *below* that expected return figure unless you want to face a meaningful risk of running out of money.

For example, if you expect your portfolio to *average* a 6% return throughout your retirement, withdrawing 6% of your portfolio in the first year and increasing the amount withdrawn each year to keep up with inflation would be setting yourself up for trouble.

The reason such a strategy is risky is that a high withdrawal rate is absolutely devastating to your portfolio if you happen to face a prolonged bear market early in retirement. After a few years of “selling low,” you’re left with too small a portfolio to benefit fully when the market *does* come back.

This poorly-timed-bear-market concept is known in finance as “sequence of returns risk.” Ignoring it completely — as the Motley Fool calculator does — will cause you to significantly underestimate the amount of savings you’ll need in order to retire.

### The Takeaways

- Don’t trust a retirement calculator unless you can see all of its assumptions and you judge them to be reasonable. One poor assumption can make an otherwise-great calculator worthless.
- Don’t overlook sequence of returns risk when planning for your retirement.

And just for fun, here’s a screenshot taken when I plugged in a 6% withdrawal rate and a 6% rate of return.

It is the nature of any model that projects events as far into the future as a retirement calculator does that the results are not 100% reliable. Unfortunately most creators of such tools make their users believe that the results are somehow magic.

Putting this together with what you’ve written elsewhere, the SPIA is starting to look like a very promising alternative.

I plugged some numbers into Vanguard, and with my birthdate of Sept. 1948, I found that $350K will buy me an inflation-adjusted annuity of $1476/monthly, or $17712/annually, $531360/30 years just multiplied out without any inflation adjustment. And $400K gives me $1687 / 20244 / 607320. Of course if those amounts are taken from a tax-referred IRA, there will be ordinary income tax, but as I will no longer have payroll taxes, I expect my overall taxes to be lower even if tax rates increase. I realize that there will be no money for my heirs, and if I drop dead at age 66 all that money is gone. But added to my estimated social security, it sounds like a pretty good deal if I survive 20+ years without major unexpected health issues. Is it too good to be true, or am I missing something?

That would be an inflation-adjusted withdrawal rate of 5.06%.

If you were to survive only 20 years, then an annuity was probably not the right way to go. (A portfolio that did nothing but keep up with inflation would support a 5% withdrawal rate over a 20-year period.) Once you start getting toward a 30-year retirement, though, a 5.06% withdrawal rate would have significant risk for a non-annuitized portfolio.

To me, if leaving money to heirs isn’t a major goal for an investor, SPIAs become extremely attractive.

That said, the question of how much of one’s portfolio to annuitize and when exactly to buy the annuity still merit careful consideration. (I know you’ve read it before, Larry, but for any other readers: I wrote a bit about those ideas here.)

The studies show that if you retired at age 65 and had a portfolio comprised of 75% stocks/25% bonds then a safe withdrawal rate has been between 3% and 4%. These studies encompass the Great Depression, WWII on up through the 60s (Businessweek’s “Equities are dead” cover) etc.

Where sequence of returns is critical is in building up the portfolio. If the good years occur in your beginning years like for those who started their first real job 20 years ago then the average return of 8% on equities has been of little help.

You might want to check out http://www.firecalc.com/

It takes your inputs tells you if your portfolio would have survived over a series of 50 year rolling periods. From the site:

“Averages don’t tell you much at all. Retire in the early 1970s, starting with $750,000 and taking out $35,000 each year, and on average you’ll do just fine. But that average is meaningless.

FIRECalc can tell you how much you would have needed to insure that you wouldn’t have depleted your portfolio if things are as bad as 1973. Or 1929. Or any of the past years for which we have data.”

It shows an example of a portfolio starting in 1973, 1974 or 1975. The 1973 portfolio first would have been depleted in 19 years, the 1974 would have lasted at least 30 and 1975 would have outlasted you easily.

So, it seems to account for the variability of returns year to year and gives an idea in what percentage of starting years your portfolio would have failed to outlive you. Pretty cool tool overall.

Mike,

I have to agree that assuming a fixed % return isn’t realistic- is there any consensus on a realistic withdrawal strategy? My thought would be something like the following:

Start with some very conservative base rate that will cover minimum expenses- 2 or 3%. If your portfolio’s inflation adjusted value is more than you started with you take out then take some fraction of the additional returns and treat it as a bonus. As long as the base rate is sustainable this strategy should be sustainable but it could mean big year to year income swings.

-Rick Francis

“…but it could mean big year to year income swings.”Yeah, exactly. A 2-3% withdrawal rate is super safe. But, for many investors, it’s darned near impossible to get your savings high enough and your fixed expenses low enough to make that a possibility.

Yeah, I like FIRECalc. It, along with Jim Otar’s “retirement optimizer” are my two favorites in that all they do are look at past returns and say what would or wouldn’t have worked rather than attempting to put a

futureprobability of success on a given plan.Can you explain how you’re getting that figure?

The monthly payments I quoted were only the initial payments, and I just multiplied out to get a sense of the minimum I would received from the annuity. If it helps, I tried a few more scenarios, all assuming an initial payment of $400K, and that I’m a single male living in NY:

With a bday in 9/48, life expectancy about 23 years according to IRS Pub 590, starting the annuity 1 year later:

- Single life only w/inflation adjustment: $1683 initial payment

- same with 20 year period: 1771

- Single only w/5% graded payment: 1424

- same with 20 year period: 1299

Same options changing bday to 9/45 to make myself 3 years older:

- 1895

- 1771

- 1498

- 1424

Given those numbers, at what point do you think it more advantageous to annuitize?

My 5.06% figure was simply the quotient of the two figures you provided: $17,712 annual payout, divided by a $350,000 premium.

As to when to annuitize, let’s stick with just one figure for comparison. (I prefer to use single life only with inflation adjustment if there’s no joint annuitant.)

So, for an investor born Sept 1, 1948, I get an inflation-adjusted payout of $1,570.83 per month on a $400,000 premium if we start the payout asap.

In contrast, if I enter a birthdate of Sept 1, 1945, and I request a monthly payment of $1,570.83, I see that I’d need to pay a premium of $357,844.

So, if interest rates don’t change between now and three years from now, you’re better off delaying annuitization for those three years if you think you can spend $1,570 per month, and still have $357,844 remaining. According to my Excel calculations, that would take an effective annual return of ~ 1.013%.

Mike,

It is good to see someone else warning about the dangers of retirement calculators. They are so highly dependent on assumptions that few people understand how to use them correctly. The result is dangerously inaccurate and misleading retirement planning. I believe this is so important that I wrote an entire ebook covering this subject.

I would even take the discussion above one step further and disagree with the high quality “backcasting” methods as mentioned in the comments above. The implied but unstated assumption behind these methods is the future will relate in some way to the past.

Given the government debt and transfer payment problems that have ballooned to unsustainable levels I think it is a fools game to assume the past will have much relevance to what we face over the next 20-40 years.

Just so I don’t come across excessively negative, the value in all these methods is they show you shades of “the truth”. Each provides a useful piece to the puzzle as long as you understand the limitations and don’t place excessive confidence in the output. That is the basic mistake most people make.

Hope that helps…