Why Most Retirement Calculators Don’t Work

New Reader? Get free regular updates from Can I Retire Yet? on saving, investing, retiring, and retirement income. New articles about twice/month. Join more than 14,000 subscribers. Unsubscribe at any time:

Most retirement calculators are broken. It’s not that the user interfaces are confusing, or the inputs are rigid and lacking, or that they have bugs — though many share those failings. No, the problems with most retirement calculators are theoretical, and go much deeper….

Let’s start with some simple division. Say you’re retiring at age 65, you’ve saved $500,000 in your 401K, and you need about $30,000/year to live on, in addition to your pension or Social Security. (These are made-up numbers: feel free to use your own.) So we divide $500,000 by $30,000/year and see that your money will last about 17 years, until you’re age 82.

That simplistic analysis sounds a little tight. What if you live longer?

Maybe we can refine the analysis. Haven’t stocks averaged about 10% growth annually? So let’s put in a growth factor: $500,000 growing at 10% annually. That’s $50,000/year in income. With only $30,000/year in living expenses, your portfolio will last forever! That’s awesome….

It’s also still wrong. For starters, we forgot about inflation. Call that 3.5%, a bit more than the historical average, considering what the government is up to right now. We’ll be lucky if it’s not worse.

We also forgot that your whole portfolio probably isn’t in stocks growing at 10%. Let’s say your combined portfolio of stocks and bonds grows at 7.5% annually. (Expecting such growth rates going forward is probably delusional, but that’s another discussion.)

So, let’s run those numbers again: $500,000 growing at 7.5% minus 3.5% (4% real return) annually, less $30,000/year in living expenses. OK, that lasts about 28 years, getting you into your early 90’s from age 65. As long as none of your forebears were centenarians, you’ll probably be all right….

Or will you?

As we’ll soon see, this simple, familiar kind of calculation gives you a bit of insight into where you stand financially today. But it tells you virtually nothing about what will happen in the future.

Even though this is essentially the same calculation you’ll find performed in the majority of the available retirement calculators, it is terribly flawed as a predictive tool. There are many reasons. But let’s start with this one: it completely ignores the sequence of returns. It assumes that average returns are all that matter.

They aren’t. You can still drown in a lake whose "average" depth is 4 feet! And you can run out of money in a stock market that averages 10% annually — if those returns don’t come until too late in your retirement.

Most standard retirement calculators assume steady growth rates, year in and year out. But that’s not how the economy or your money works. When you are adding to, and especially when you are withdrawing from, a portfolio, deeper analysis shows that you will wind up with dramatically different sums, depending on the order in which returns, or losses, accumulate.

Some calculators try to model the fluctuations of the markets by incorporating Monte Carlo simulations into their calculations. (The name is a reference to the famous European casino.) These simulations require you to input a range of possible values for every parameter (inflation, investment returns, and so on). The calculator then proceeds to combine these parameters into hundreds, thousands, even millions, of random variations. Surely that will cover all the possibilities? Well, the output data is definitely impressive: You may have seen one of the resulting net worth versus time graphs, bursting with scenario lines, a veritable financial Medusa….

These Monte Carlo simulations do have some value in showing you the outlying envelope of possible paths to your future. But the verbosity hides a fatal flaw: the vast majority of those paths are very unlikely!

It turns out that markets aren’t actually completely random, and the randomness they do exhibit is not the normal Gaussian randomness most often used in Monte Carlo calculators. Markets usually operate in cycles of varying lengths, unpredictable at first, but eventually recognizable. So stacking a bunch of totally random factors on top of each other doesn’t make for an accurate simulation. The fatal flaw with these calculators then is that they introduce too much randomness, ignoring long-term trends.

So if steady growth and Monte Carlo-based calculators are misleading, what about using actual market history? Some leading financial thinkers have argued that this is the most realistic way to simulate retirements. After all, if your portfolio could have survived every single historical event of the past 100 years, shouldn’t you be able to sleep easy about your upcoming 30-40 year retirement? FIRECalc has popularized this approach, and it has many strong adherents because of the generally conservative results.

An approach to retirement planning based on market history is compelling, and better than many of the alternatives, but it still has serious flaws:

  1. It assumes that the future will be like the past.
  2. It ignores current market valuations.

History might rhyme, but it doesn’t precisely repeat itself. If we have no other choice, we can plan based on past experience. But, somewhere deep inside, we know that the future will turn out differently. The modern world, with its complex technological, financial, and social systems, has the potential for great prosperity, or for great disaster. Who can really say we’ve already seen the best or worst of it?

Practically speaking, if you’re going to predict the future based on the past, you should at least take into account your starting point, and that’s why market valuations matter. Researcher Wade Pfau has used historical return data to show that safe withdrawal rates for portfolios are highly correlated to the level of the stock market at the time of retirement. Given the current historically-unprecedented economic conditions, he has suggested using bond and stock returns that are 2-3% lower than past averages.

So that’s it. We’ve seen that most common kinds of retirement calculators are flawed in one respect or another. But we shouldn’t be so shocked or disappointed. In essence they are all trying to predict the future. And we all know the track record for crystal balls….

What to do about it?

You can continue to chase more and more sophisticated calculators. And I’ll probably do a little of that here. But it will be with a healthy dose of realism.

Frankly, I expect those more sophisticated algorithms to converge on the same conservative result: If you want to be absolutely certain you don’t run out of money over the course of a lengthy retirement, you’d better consume no more than 2-3% of your assets each year early on.

In the end, no retirement calculator can predict the future. All it can tell you is which direction you’re headed. But the better calculators are worth paying attention to for this reason: they’ll give you enough warning to make course corrections before you drive over a cliff — run out of money, that is.

A retirement calculator can tell you which direction to steer, but it can’t really tell you with any precision where or when you’ll arrive….


  1. The frustrating aspect of eliminating the final 1% of a retirement calculator's failure scenarios is that it boosts the probability that we'll have more money than we need. That implies working longer than we want or spending less than we could. And 90% of the time, it'll be way more money than we need.

    Thank goodness for the retirement scenarios that calculators don't handle very well yet: variable spending plans and annuities.

  2. Darrow Kirkpatrick says:

    Good points. As in so many areas of life, we can move to the extremes (and possibly wind up in the ditch), or we can stay in the middle of the road (and be prepared for evasive action). Thanks Nords!

  3. Freesia blossom says:

    What are some of the better calculators? I've tried the one AARP has, but I'm unsure of what is assumed in the background. For example, does it assume one already owns, out right, their own residence by retirement and those costs are minimized in their calculator results of what is needed in retirement? What is good about it is that you can readjust the expected percentages of returns on investment (their default percentage is pretty optimistic).

  4. Darrow Kirkpatrick says:

    Hi Freesia. You're right to dig into the assumptions behind any retirement calculator you use. I agree, the AARP doesn't seem to make those assumptions very obvious, at least that I can see easily. Some calculators will include a link to the "fine print" that explains how their model works. That's a minimum I'd want, before I'd put much stock in one.

    The three best retirement calculators that I've used are from Schwab, E*Trade, and FinancialMentor. However I can't really recommend any single calculator. Almost all of them have to make assumptions (or take your assumptions) about the future, and that's a fundamental limitation. It's best to run several calculators, and treat the results as guidance, not gospel.

    In general, if a retirement calculator doesn't ask about your home, I'd assume that when you input your cost of living, that number should account for your shelter cost. So if you didn't own your home outright, that cost of living would need to include mortgage interest or rent payments.

    Hope this helps. Thanks for the comment!

  5. Nice article but you missed one huge point. If I knew the exact date of my demise, then any of these calculators will work. Trouble is your own death is the biggest variable in the whole calculation. It could be off by 50% or more. Why quibble a few percentage points over inflation, bond rates, stock market performance? As in life, death is a mystery.

  6. Darrow Kirkpatrick says:

    Sure, good point. Add life expectancy to the list of unpredictable variables required to do an accurate retirement income calculation. To be fair, the converged conservative result that I mention above — consuming no more than 2-3% of your assets each year — may not require a life expectancy in the calculation, because the principal is likely to last indefinitely. Thanks dh.