Richard Connor

**I RECENTLY SPOKE** with a colleague. I’d expected him to be retired by now. He told me that he’d planned to retire last spring, but his employer offered him a three-day-a-week part-time schedule with full benefits. He discussed it with his financial planner.

The planner told him that, if he retired, he had an 85% chance of meeting his retirement goals. By working part-time for two more years, his chances of meeting his goals went up to 95%. My colleague enjoyed his work and figured he’d still be able to schedule four-day weekends. He decided to continue working part-time.

If you work with a financial planner, you may have heard assessments similar to my friend’s. The higher the percent chance of success, the better you’re supposed to feel. But what do these numbers really mean?

There are two common methods that planners use to assess your portfolio’s probability of success—historical averages and Monte Carlo simulations. Using historical averages is more straightforward. You take historical market returns for each asset class and compute a weighted average annual return based on your portfolio’s asset allocation.

Let’s say you have a classic 60% stock-40% bond mix. The stock portion is in an S&P 500 index fund, which has an historical annual return of 9.6%. The bond return is calculated using a long-term U.S. bond fund, which has an historical annual return of 5.6%. Their weighted average is (0.6 x 9.6%) + (0.4 x 5.6%), which comes to 8%.

Starting with this assumption, a retirement income projection is built for each year of retirement. It compares all sources of income—pension, Social Security, portfolio growth, withdrawals—against all anticipated expenses. The analysis can incorporate other variables, such as inflation, required minimum distributions and taxes.

If your total spending is less than total income, your portfolio continues to grow. If you have money left over at the end of your whole retirement period, your retirement plan is considered a success.

What if your spending is greater than your income? The analysis assumes you make up the difference by taking larger withdrawals from your portfolio. Your portfolio will shrink each year and may eventually be depleted. If your savings are projected to run dry before the assumed end of your retirement, your retirement plan is—needless to say—considered a failure.

The weakness of this analysis: The model uses the same investment return for each year of the analysis. Your retirement success will hinge on how much your actual investment results differ from the assumed average return.

A Monte Carlo simulation attempts to avoid the shortcomings of the “historical averages” method by varying the returns of each asset class in each time period, often focusing on annual results. Once again, you start with the historical averages for each asset class. But a Monte Carlo simulation throws in historical volatility and a random number generator, thereby calculating a distinct asset return for each time period in the retirement projection.

The Monte Carlo method repeats the whole process hundreds or even thousands of times, each with a different set of returns for each asset class and each time period. The total number of iterations that produce a positive portfolio value at death is computed. This number is then compared to the total number of iterations attempted.

Say the Monte Carlo software ran 1,000 iterations of a retirement projection. If 950 of those cases showed a positive portfolio value at the end of the projection, the probability of success for that retirement plan would be 95%. In other words, each of the 1,000 scenarios considered a different stock and bond market performance and, in 95% of those random cases, your portfolio carried you all the way through your retirement.

The math is more complex than my explanation here. You can also use the statistical methods to randomize more than asset returns. For example, you could vary inflation randomly as well. But varying asset returns suffices for many of my friends and colleagues who aren’t math aficionados.

What if this analysis shows a lower probability of success than you’re comfortable with? You can often adjust your plan to reach an acceptable level. For example, you could reduce retirement spending. Alternatively, you might adjust your asset allocation, increasing risk until you meet your success objective. Just don’t overdo that extra risk.

Michael Kitces has written a paper that challenges the need for a 95% probability of success. The paper brings up several good points that can help us better understand the results of these experiments.

Let’s say you only achieve a 50% probability of success. Small adjustments may translate into a dramatic improvement. For instance, if you have ongoing sources of income, like pension and Social Security, that cover the great majority of your spending, cutting your spending just a bit could significantly improve the odds of success.

The bottom line: If you’re working with a financial planner and she says your probability of success is 50%, don’t panic. It’s likely not a bad starting point as you approach retirement. The key is to understand the source and magnitude of the failure, and what can be done to overcome it. Consider my friend at the start of this article: A few years of part-time work put him squarely in the 95% success range.

*Richard Connor is **a semi-retired aerospace engineer with a keen interest in finance. He** enjoys a wide variety of other interests, including chasing grandkids, space, sports, travel, winemaking and reading. **Follow Rick on Twitter *@RConnor609* and check out his earlier articles.*

*Do you enjoy HumbleDollar? Please support our work with a donation. Want to receive daily email alerts about new articles? Click here. How about getting our newsletter? Sign up now.*

DI recently had a fee-for-service CFP run simulations for me. I had him run them for 2%, 4%, and 5% inflation. I have a Plan B which I will need if inflation stays at 5%. Although I have been retired for twenty years I have not needed to draw on my portfolio for more than a few thousand a year a few times, but I will need it when I move to a CCRC in two years. I am aware of SORR, but I have enough in cash and short term bonds to last for a while.

If investments contribute a large portion toward the likelihood of meeting long-term retirement goals, then sequence-of-returns risk in the early retirement years is a key planning consideration. The 30-year retirement outcome can be vastly different if the stock market happens to halve or double soon after retirement. Likewise, 1980’s-like inflation could wreck havoc on retirement plans.

I think retirees are always well advised to build some “fat” into retirement plans just like your friend.

There is a better “historical analysis” calculator that I use. It’s called FIRECalc and I like it because it does not rely on historical averages. Instead, you pick a duration for your retirement. Let’s say 35 years. You plug in your starting balance and your annual spending. The calculator evaluates every 35 year period since 1871. So it would tell you how your portfolio would have fared from 1871-1906, and each subsequent 35-year period through 1985-2020.

You can also play with advanced functions, such as inputs from expected social security income, lump sum withdrawals/additions, asset allocation, etc. In the end, it spits out a probability of success evaluating all of the time periods available, much like any other calculator.

I feel much more confident in this style calculator, opposed to one that simply relies on a historical average. I also like Monte Carlo simulations, though their “randomness” may generate some unrealistic (or at least unprecedented) scenarios which would pull down your probability of success.

Brent, thanks for the review and helpful reply. I’m familiar with the historical time series technique and have seen it used by researchers such as Wade Pfau. I agree it is superior to the historical average technique. I thought about including it but chose not to due to length concerns, and my primary purpose to explain the Monte Carlo approach my friend was presented.

I haven’t looked at FIRECalc in a few years. I’ll have to go back and review it. There are a number of free and for purchase retirement calculators available. I’m thinking of reviewing a few for future articles.

I’ve also used FireCalc (as well as Fidelity’s Retirement Planner and Quicken’s Lifetime Planner, as well as a few others). Highly recommended.

FireCalc is quite nuanced in its inputs and the chart output of the historical results of a particular withdrawal rate, etc is quite helpful. If you have some margin and can adjust (less travel, etc), I suspect 75-80% is probably more than good.

Since semi-retiring 7 years ago and fully retiring 2 years ago, my “success rate” based on past historical results has gone up to 100%. Thank you markets! That said, SORR (or the risk of markets collapsing for a lengthy period right after you retire) is real, and FireCalc does display this.

Rick, a review of FIRECalc and other free retirement calculators would be great.

How does this all work if you base withdrawals just on the spin-off of dividends, and interest and perhaps mutual fund capital gain distributions?

May not be practical for total retirement income, but as supplement to SS.

Good question Richard.As you suggest I think it is more applicable to a Total Withdrawal approach. I look at Monte Carlo (and Historical time Series analysis that Brent mentions above) as a way to demonstrate how big a margin of safety your plan has. Very high scores give you confidence, but I would still watch your yearly performance. If someone’s plan allows them to live off D&I, I would think they have a lot of margin.