Monday, October 12, 2009

When Can I Retire?

Happy Thanksgiving, Canada! This is a Thanksgiving version of the usual Sunday feature looking back at selected articles from the early days of this blog before readership had ramped up. Enjoy.

A major concern for many people is how old they will be when they can retire. This depends on a number of factors such as how much you save, how well your investments perform, how much you spend during retirement, and how long you live.

Retirement calculators can figure this out for you based on a number of assumptions. However, most of them don’t give you a feel for how the final answer would change if your investment returns are volatile instead of perfectly steady.

There was a good post over at the Canadian Financial DIY blog about using Monte Carlo analysis for financial planning. Monte Carlo analysis just means simulating possible outcomes many times to see how the final answer changes.

I decided to use Monte Carlo to see how the mix of stocks and bonds in a portfolio affects when you can retire. I had to make some assumptions:

- Stocks and bonds will have the returns and volatility as reported in the paper Portfolio Optimization by John Norstad (2002-09-11).
- Retirement money has to last until you are 90 years old.
- After retirement you will invest conservatively and get steady returns 3% above inflation.

The simulations were based on the following saving and spending rates:

- You save $600 per month (rising with inflation) in a retirement account while working.
- You will spend $4000 per month (in today’s dollars) during retirement.

I ran each simulation until there was enough money to retire. After a million runs, I found the median retirement age as well as the age range that covered 90% of the simulations.

Here are the results if you start saving at age 25:

StocksBondsRetirement Age
Median (Range)
0%100%75 (65-82)
50% 50%66 (56-76)
100%0%57 (47-72)

The width of these age ranges shows that there is quite a bit of uncertainty. There isn’t much doubt that historically stocks have been better than bonds at giving investors a chance to retire early.

Here are the results if you waited until age 40 to start saving:

StocksBondsRetirement Age
Median (Range)
0%100%80 (74-84)
50% 50%75 (68-81)
100%0%69 (60-80)

The advantage of stocks is evident here as well. The other thing to note is the power of starting to save earlier. In the all-stock case, the 15-year head start in saving led to retirement about a decade earlier.

It makes sense to use short-term bonds for money that you will need to spend within the next 3 years or so, but I haven’t found a good reason to own bonds for the long term.


  1. Hi Michael. I would have agreed until a recent re-post of yours showed the tremendous value of reducing volatility. The usual rationale for bonds is that they are negatively correlated with stocks in the short term, which should have a powerful calming effect on volatility. Did your simulations here take that into account?

  2. Patrick: The data I used for the simulation was from a paper by John Norstad (see the post for a link). The long-term correlation between stocks and bonds is actually positive (but very small). Including bonds in a portfolio does reduce volatility significantly, but it also reduces the expected return. The important return is the compounded return, which is the expected return less a volatility penalty. Unfortunately, the volatility benefit of owning bonds is more than offset by the expected return penalty of owning those bonds.

  3. This post scratches the surface of retirement planning, which we all know is more complex than could be covered in a single post.

    It's worrying to see that the retirement ages in the examples are quite late in life. Does it assume people are able to match the market? If so, the reality of people under performing the index makes an even less optimistic case.

    Of course there are countless other assumptions to fiddle with, but let's keep things relatively simple.

  4. Gene: My simulations assume that the investor buys and holds. So, that means matching the market without paying high MERs or underperforming due to panic selling, etc.

    Any simulation that shows it is possible for the majority of people to be able to retire into a middle class life for decades is suspect. There aren't enough workers to do all the work to give retirees the lifestyle they would want.

  5. This comment has been removed by a blog administrator.

    1. Here is Patrick's comment with a broken link removed:

      Hi Michael,

      According to [a chart whose link no longer works], the correlation between AGG (iShares Bond ETF) and VV (Vanguard Large-cap US stocks ETF) is -0.34.

      This strains credibility, but if it's true, the Norstad paper gets a whole lot more interesting.

  6. Patrick: I had the same confusion when I first looked at this web site. The correlation between AGG and VV is -0.34, but this is over only the last 3 months. Over the last 2 years this correlation is -0.06. To be interesting, we need correlations over a much longer period of time. If someone were able to predict the correlations for the next 3 months or 2 years, that would be powerful information.