I read a lot of websites that say, "I've run the numbers, and you should be saving around X% of your income for retirement in order to be safe, and have fruitful golden years." With X% being a number at or around 15%.

Getting to their assumptions when they "ran the numbers" has proved to be nigh impossible. They also seem to be limited to items like, you make $50,000 per year and your investments return 10%. There has to be more underlying assumptions, therefore, I am asking:

What assumptions are tacit in the statement that "saving and properly investing 15% of one's income over a lifetime is a pathway to a successful retirement?"

By this, I mean items along the lines of:

  • Single, married, or single and dating at time of retirement?
  • Retire at 55, 60, 65, 80?
  • 25K/yr income, 50K/yr income, 150K/yr income?
  • That 15% goes to a tax advantaged account? (i.e. 401k, IRA, Roth, your nephew's 529?)
  • Kids or no kids?
  • Paying for said kids' college or not paying?
  • 2003 to 2007 returns, 2008 to 2009 returns, or "I averaged the whole S&P500 over 2 world wars and order of magnitude technology advances" returns?
  • Dual incomes the whole time?
  • Making a lot more money at time of retirement, or making about the same money as early to mid career at or about the time of retirement?
  • Renting a house the whole time, or owning a house as early as possible?
  • Obscene ROR's on that house, or assuming it loses money?
  • Your expenses go down at retirement, stay the same, or go up?
  • Medicare exists? Social security exists? Tax rates go up, down, stay the same?
  • You'll never get laid off, you might get laid off, you get laid off often?
  • Family helps you out with major purchases, or does not?
  • No, big, or modest inheritance?
  • Live in a cheap area, or live in a "statistically average for costs, land, CPI, wages" area of the US?
  • Taxes go up, taxes go down, taxes stay about the same?
  • Leaving a nest egg when you die? Or, dying broke?
  • Living to the statistical average age of men and women in the US, or, living to be 101.2?
  • Inflation under control? Inflation to the moon? Has it considered deleveraging and deflation?
  • State with an income tax? Or only a state with sales tax?

I like to run little what if spreadsheets for retirement and every time I do, 15% comes in woefully inadequate for retirement. This surprises me, as, I consider my life plan to be among the easier (read, cheaper) ones to find a tenable retirement solution. I.e. age 31, no kids, no plan to ever have kids, no debt, home buying adverse, upper income per capita job, eventual wife will likely work full time, extremely mobile and willing to move, and with a job skill set that is likely to not be out of work often.

Edit: Assumptions that usually land me in hot water are: long term rates at 4% to 5%, salary adjustments of ~4% per year up to a cap (a cap equal to what a senior person in my industry is paid, has mimicked my salary raises surprisingly well actually), I assume a 20% tax rate on earnings averaged over all accounts, then I seek to replace an "inflation" adjusted 100K at ~1.5% per year (my real goal would be a CPI adjusted 100K into the future, which very likely would not be driven by inflation, but no one has one of those crystal balls).

Using those, 15% doesn't cut mustard. Which makes sense since most estimates below seem to boil down to replace $X salary at age Y assuming Z% ROR.

  • 2
    Maybe you could share some more information on your simulations. What assumptions are you making? What kind of lifestyle do you plan to live in your retirement years?
    – Jason R
    Commented Mar 16, 2012 at 16:48
  • Added my "typical" assumptions. Commented Mar 17, 2012 at 17:49
  • 5
    This is a refreshingly-complete, well-thought-out question with many assumptions that many people don't generally consider when reading things like 'save 15% of your income' -- +1
    – johnny
    Commented Apr 6, 2012 at 22:58
  • It assumes you will live to retirement age in the first place. Commented Apr 18, 2018 at 21:29

6 Answers 6


There's no magic. Usually these models set out to replace 60-65% of your gross income in retirement.

For example, if you:

  • Start at $50,000 / year, and average 2% raises each year for 30 years (Final salary about $90k)
  • Contribute 15% to retirement
  • Yield a 7% annual return on investment (calculators will usually vary between 4-8% annual return)

You'll retire with about $850k. That will let you generate an income stream of around 55k for your expected lifespan.

Is 15% the right answer for you? No idea -- it depends on what you want, how you invest, and what you can afford.

  • 1
    +1 -- Great answer, The only problem with this type of analysis is that it mostly ignores risk of all types (specifically investment return risk and length of lifetime risk). I have been really intersted in Monte Carlo analysis for this reason.
    – Pablitorun
    Commented Mar 17, 2012 at 2:03
  • 1
    I actually think this gets to the heart of the question. There is no magic, as Duffbeer703 says. There are no deep assumptions, just a replacement goal and an assumed ROR. Commented Mar 17, 2012 at 17:45

I wrote Retirement Savings Ratio some time ago and 15% seemed the sweet spot. The 15% is 10% saved and 5% match, in my mind. The reason I link is to show a spreadsheet that shares my approach to the analysis.

You list a huge number of things to think about, all great. I'm near Duff's view although I shoot for 20X final income so 4% of that is an 80% replacement. There's nothing magic about 80. If social security exists, it actually drops the required 80 from you to the 60 or so Duff lists, so we align.

The 'problems' with my sheet - no one gets exact 3% raises for 43 years, so one can go in and change that, maybe higher early on, then slow it down. Returns are 8% every year, so one must go to the sheet or their own plan and reassess how they are doing each year. Disclosure - we are now at 14X our income with a goal of 25 to replace 100%. So we need a cumulative 80% growth (as in S&P at 2500) or the years of deposits to get there. I am nearing 50.

Let me pick one of your questions - "Your expenses go down at retirement, stay the same, or go up?" - this is the $64K question. the 80% is an average guess. The general assumption is in this range. If we downsized right after the kid goes to college, we could retire on far less. So, to wrap up, online calculators and spreadsheets only go so far. Your questions are dead on. A planner should ask every one of these questions before assuming any ratio.

Edit - on re-read I am blown away by the list of questions you have here. It's great. And it shows how there's risk no matter what.

  • 3
    In regards to expenses go up or down at retirement, some expenses may go down like downsizing, having paid off the mortgage, kids left the nest, etc..., however other expenses may go up, you will have more free time on your hands so if you are still active you may tend to go out alot more and take more holidays just as examples. So these things need to be looked at, and you need to ask yourself: 'What type of lifestyle would I want in retirement?', and then start planning for it early on.
    – Victor
    Commented Mar 16, 2012 at 22:29
  • +1 -- Great answer +2 if I could, My only problem is that this sort of static analysis doesn't do a great job with risk. This is your whole 'problems' section of your answer. I have been really interested in Monte Carlo analysis as a way to answer that question with more definitive answers.
    – Pablitorun
    Commented Mar 17, 2012 at 2:05
  • 1
    @Pablitorun - absolutely right. I've seen Monte Carlo in action and it adds value to the analysis. The only defense I'd have for my spreadsheet is (a) free, (b) the user is welcome to tinker, don't like my 8% return? change it up or down. / With the rate of financial illiteracy, this is a vast improvement over nothing, but a visit with a pro with the heavy duty tools can't hurt. Commented Mar 17, 2012 at 3:12
  • Great answer indeed. I think I will take the base premise that Joe has setup, and see if I can't add some probability to it. Commented Mar 17, 2012 at 17:52
  • for sure your spreadsheet doesn't need any defense. Monte Carlo analysis is well beyond the scope of the average investor. (Simple percentages are beyond the scope of the average investor IMHO.) I only mentioned it here as 1.) @GlorifiedPlumber seems plenty competent, and 2.) I thought you might be interested.
    – Pablitorun
    Commented Mar 19, 2012 at 1:34

I argued for a 15% rule of thumb here: Saving for retirement: How much is enough?

Though if you'll let me, I'd refine the argument to: use a rule of thumb to set your minimum savings, then use Monte Carlo to stress-test and look at any special circumstances, and make a case to save more.

You're right that the rule of thumb bakes in tons of assumptions (great list btw). A typical 15%-works scenario could include:

  • starting to save at a young age and retiring 30-40 years later
  • decent but not pure-fantasy investment returns
  • relying on social security's existence
  • your income is low enough that social security isn't lost in the noise
  • you are willing to spend down to zero and leave no inheritance

If any of those big assumptions don't apply to you (or you don't want to rely on them) you'd have to re-evaluate.

It sounds like you're assuming 4-5% investment returns? As you say that's probably the big difference, 4-5% is lower than most would assume. 6-7% (real return) is maybe a middle-of-the-road assumption and 8% is maybe an unrealistic one.

Many of the assumptions you list (such as married/kids, cost of living, spouse's income, paying for college) can maybe be bundled up into one assumption (percentage of income you will spend). Set a percentage budget and as you go along, stay within your means by sacrificing as required. Also smooth out income across layoffs and things by having an emergency fund. By staying on-budget as you go you can remove some of the unpredictability.

The reason I think the rule of thumb is still good, despite the assumptions, is that I don't think a "more accurate" number based on a lot of unpredictable guesses is really better; and it may even be harmful if you use it to justify saving less, or even if you use it to save far too much. See also http://en.wikipedia.org/wiki/Precision_bias

Many (most?) important assumptions are not predictable: investment returns, health care inflation, personal health, lifestyle creep (changing spending needs/desires), irrational investment behavior.

I agree with you that for many scenarios and people, 15% will not be enough, though it's a whole lot more than most save already. In particular, low investment returns over your time horizon will make 15% insufficient, and some argue that low investment returns over the coming 30 years are likely. Without a doubt, 20% or more is safer than 15%.

Do consider that "saving enough" is not a binary thing. If you save only 15% and it turns out that doesn't completely replace your income, it's not like you're out on the street; you might have to retire a few years later, or downsize your house, or something, but perhaps that isn't a catastrophe. There's a very personal question about how much to sacrifice now for less risk of sacrifice in the future.

Maybe I'd better qualify "not a binary thing": some savings rates (certainly, anything less than 10%), make major sacrifices pretty likely... so in that sense there is a binary distinction between "plausible plan" and "denial."

Also, precise assumptions and calculations get a lot more useful as you approach retirement age. You can pretty much answer the question "is it reasonable to retire right now?" or "could I retire in 5 years?" (though with a retirement that could last 30 years, plenty of unknowns will remain even then).

I think at age 20 or 30 though, just saving 15% (20% if you're conservative), and not spending too much time on a speculative analysis would be a sound decision. That's why I like the rule of thumb. Analysis paralysis (saving nothing or near-nothing) is the real danger early in one's career. Any plausible percentage is fine as long as you save.

As your life unfolds and you see what happens, you can refine and correct, adjusting your savings rate, moving your retirement age around, spending a little less or more. The important thing earlier in life is to just get in the right ballpark.

  • 2
    So I think Duff's gets to the heart of my question, namely "there are no assumptions." But I wanted to weigh in on this answer specifically because I think it is really good. You make a phenominal point, I can make an enormous list of assumptions, and attempt to disposition them in any model I do, but that doesn't necessarily do anything towards actually generating a "more useful" or "accurate" answer. I had not considered precision bias in my quest for accuracy. Commented Mar 23, 2012 at 3:14

duffbeer and JoeTaxpayer give great answers, and you should probably accept one or the other, but I would like to recommend trying some of the online tools for Monte Carlo analysis.

A Monte Carlo analysis is essentially plugging in a range of possible values (a probability function) for yearly values of pretty much anything involved in your financial life: salary growth, investment rate of return, expected life span, etc, etc, etc....and then running thousands of simulations on those values to give you the probability that your money will last until you die. It is basically a really souped up version of a "what-if" spreadsheet.

A google search for "monte carlo retirement planning" may sound like a hilariously bad idea, but it lists some free tools that you might find interesting or useful. I myself have only used the product from Financial Engines quite a bit, but I believe that is only available through your employer. (If they offer it.)

  • 1
    +1 for Monte Carlo. I added a link to an online one that seems interesting. It's a valuable tool. Commented Mar 17, 2012 at 3:24
  • +1 also. I've done this before, but on a much lesser level, just assigning random probabilities to ROR, inflation, and salary adjustments that year. The concept of expanding it is intriguing to me, I will see what software the company has access to. Commented Mar 17, 2012 at 17:28
  • This is a really interesting way to validate your assumptions and assess risk! Commented Mar 19, 2012 at 14:35
  • 2
    Couldn't be worse than my "Las Vegas retirement planning"
    – JohnFx
    Commented May 31, 2012 at 18:10
  • 1
    @JohnFx Also much better than "Martingale retirement planning"
    – user12515
    Commented Sep 28, 2020 at 3:24

Some of the other answers have commented on this, but I particularly want to comment on the "Your expenses go down at retirement, stay the same, or go up?" assumption people often make. We assume we'll spend less in retirement, but as David Chilton points out in "The Wealthy Barber", "it's a myth that present-day retired people spend much less money than their working counterparts." He cites that reduced expenses like paid-off mortgages and children leaving the home usually happen several years before retirement, so "the last few years at work are characterized by good income levels and significantly reduced expenses. In all likelihood, disposable income is at its highest point ever." Furthermore, some expenses go up, e.g. golfing, hobbies, traveling, all of which cost money. Healthcare costs often increase as well, and Chilton goes on to discuss "a retiree's number one enemy" — Inflation. So in my mind, the answer to that question/assumption would be "costs [will probably] go up," and 60-80% of retirement income won't cut it.


I think your very long list of possible assumptions makes a tacit point of your own: to state "15%" as a general value is bogus. I think, in most cases, the "15%" is merely a popular meme.

To give any fixed number or percentage of income saved is insufficient without expanding things in the way you show. Therefore, a formula, in which at least a handful of variables can be plugged into it, seems like the right approach. (And this is what is being discussed here with the Monte Carlo method).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .