Monte Carlo

I have been playing with ESPlanner for the better part of a month, and like Dan Royer, I have come to rely primarily on the simple “economics-based planning” method, even though the Monte Carlo capability was an attractive point in choosing the software. The “economics-based planning” is easier to understand, and of course takes much less time to generate reports. I am close to retirement (I hope), and going forward use 1% to 2% real return estimates, partly to reflect my feeling that returns over the next couple of decades will be less rosy than history would suggest, but mainly to cushion sequence-of-returns risk.

Nevertheless, it would be nice at some point, after I settle on suitably pessimistic expectations for taxes, social security benefits, etc., and refine the balance between qualified and non-qualified asset spending (using “key ages” choices, etc.), to look at the Monte Carlo stuff. However, I see a couple of problems with this.

First, I think the defined asset classes incorporate too optimistic an outlook (not your fault, they’re just built from history). It appears that the only way to define an asset class with different mean returns, etc., is to input ten years or more of data, and let ESPlanner then tell me what I’ve got. After a lot of experimentation, I am not making much progress on, for example, defining a modified large cap equity asset class that is similar to the ESPlanner-supplied one, but with a lower mean return. I gather from questions in this forum in previous years that it used to be possible to define an asset class by inputting mean, variance, etc., but this was taken away in subsequent versions of the software. I assume you had some good reason for removing this capability? Or do I misunderstand?

Secondly, when running a Monte Carlo analysis, one is offered a choice of three (somewhat arbitrary, but on the whole quite reasonable) spending levels, then given a set of tables that one can study to get a sense of how likely one is to be able to maintain the chosen spending level. It is a rich set of data, but sometimes a little hard to interpret, at least at first glance.

Other Monte Carlo calculators with which I am familiar mostly (and simplistically) let you specify a level of spending, initial assets, and a time frame, then give you the probability that that level of spending can be sustained until death – e.g. 85% likelihood of success, or 15% likelihood of ruin. Of course, these calculators allow none of the complexity of input that makes ESPlanner so great. But I wonder, would it be possible at some point to add that capability to ESPlanner? I realize ESPlanner provides far more, but it would be nice to be able to get a “quick and dirty” idea of success, without studying the tables. Or, given ESPlanner’s multiple inputs, and focus on smoothing, is this not even theoretically possible within the established framework of the program? Or is it theoretically possible, but too difficult to program? Or is it easy to program, but with the result that ESPlanner has to run for impractically long times to produce results?


Good feedback on Monte Carlo. For your first issue, the last major release made the switch to inputting a sequence of returns instead of the old way. The simplest response is that user defined assets (UDAs) have been a challenge for years and this change was the best way to address various issues. For example, on the old version, if you used odd combinations of inputs there was a chance of getting outputs that were distorted. Also, changes to make the results more meaningful and easier to understand is definitely a good thing.

I don't want to distract from your other comments, but it seems that overall you are really getting at how to do risk management within ESPlanner. Here are some ideas that I've used.

To cope with uncertainty and risk, users can take a number of steps including:

1. Researching and setting "reasonable and conservative" assumptions throughout ESPlanner inputs.
2. Sensitivity testing to determine which choices (or combinations of choices) result in positive or negative outcomes along with increasing / decreasing risk.
3. Developing a thoughtful contingent plan.
4. Explicitly including healthcare and potentially long-term care inputs (e.g. through special receipts / expenditures and contingent plans) in your profile.
5. Modeling “what if” scenarios (e.g. increase / decrease assets, higher / lower rates of returns, inflation, potential tax increases and Social Security cuts, temporary job loss, etc.). Some of these can hit in combination and understanding your personal range of “reasonable” ESPlanner estimates can help you understand your downside risk and margin of safety along with other factors.
6. Various forms of insurance including life insurance.
7. Using the "safety factor" approach (in the next comment below) or other methods including reserve funds, bequests, estates, special expenditures / receipts, etc. to build a risk “cushion”.
8. Understanding which safety cushions are implemented in your profile. These can be modeled to estimate how large a cushion you have along with opportunity costs.
9. Setting Monte Carlo (MC) spending behavior to cautious or conservative. However, the limited data underlying many MC portfolios leads to additional risks as future events may be very different from historical returns and correlations. Upside Investing fits here as well.
10. Adjusting real life financial, health, employment, housing, insurance, etc. decisions and associated actions.

Overall, depending on your vulnerabilities, you can try to adjust and manage those areas that seem most critical.


For reference, here is the "safety factor" approach I mentioned above.

Spending less than ESPlanner’s recommendations can be easily modeled using a “safety factor” approach based on the “Standard of Living” (SOL) index in the Assumptions tab. To do this, create a copy of your profile and set your last few years to a high SOL. This forces the program to reduce consumption in earlier years and preserves assets for late in life. Essentially, you are shifting potential consumption to the end of your life (or the life of your spouse). This can be used to create a margin of safety, gifts or bequest, as a resource for late in life healthcare and/or long-term care or other reasons.

The beauty of this approach is that it preserves all of the year by year taxes, spending, asset levels, etc. that are part of ESPlanner’s calculations and reports so you do not have to guesstimate them with side calculations. This can also be used to create different safety factors ($ or %) at different points in time.

The approach is very flexible and only takes a few minutes to fine tune. For example, say you want to spend 10% less than ESPlanner’s “ceiling” for the next 2 years, then 5% less until end of life. To do this, keep the Standard of Living at 100 for 2 years, then raise it slightly to say 110 for several years, then raise the SOL again to 200 or higher during the last few years of life. The exact numbers will vary depending on your specific profile, but this should get you in the right neighborhood. Each year, you can adjust the safety factor again as you wish.



I like the SOL safety factor you describe; it hadn't occurred to me. It looks at first glance to be about the "cleanest" way of creating a cushion.

That's how I see it. I've tried every other way I can think of and it works great for me.


Nice. It definitely is better than inputting a bunch of special expenditures - especially if you want to make changes later!

Doesn't specifying a bequest accomplish a safety factor also? Seems simpler.

It is simpler, but can cause other issues such as unnecessary life insurance recommendations which may skew the results. Also, the safety factor approach can let you set the amount at any level you decide so has some additional benefits.


We use cookies to deliver the best user experience and improve our site.