Hi Guys,
A few days ago Catch22 posted a request for a little help in constructing a portfolio for a retiring couple. The response was huge, literally a tidal wave of informed questions and excellent suggestions. That was somewhat surprising given the fact that the profile for the retiring couple indicated that they were relatively well healed, and, for the most part, had pretty much all their ducks in proper alignment.
This was not a problematic assignment, yet the enthusiasm was infectious. Retirement planning occupies every investors planning process at least one time. It is one of the seminal events in a lifetime. The decision itself and the decision making process are stressful but necessary exercises.
Although decision making is more art then science, most retirement planning experts favor examining multiple options and doing “what if” scenario drills. That’s because the future is so uncertain. The decision to finally pull the retirement trigger is often painful. Sometimes analysis paralysis adds to the discomfort. The saving news is that there are some nice resources nearby on the Internet.
The mathematical tool that is specifically designed to address uncertain outcomes is Monte Carlo simulations.
All the major mutual fund houses acknowledge the retirement decision tipping point and the mental anguish it precipitates. They have reacted with free excellent Monte Carlo-like planning tools. That’s good.
I know, I know you’re saying” there he goes again”. That’s true. But within the last month I discovered a “better” Monte Carlo tool. I promise this is the last such posting (well at least for a few weeks).
Some investors are predisposed against statistical analyses, especially Monte Carlo techniques. It is perceived as far too mathematical, too exotic, too sophisticated. Nonsense; you need not know how to build a car to use it. There is financial risk to such ruinous behavior. The mathematics and the random selection of parameters is not conceptually complex; it is quite simple.
If that’s true you might ask, then why is the method not more commonly applied? The answer is that it is, especially since the proliferation of the home computer.
The speed of the modern computer allows the simple procedure to be executed thousands of times while a labor intensive pencil-and-paper approach could only evaluate a single scenario. The particular code that I will recommend does 10,000 random cases for each situation specified. Decision making teachers all endorse multiple option explorations over limited examinations. That’s the beauty and primary advantage of Monte Carlo simulations.
There is a large and constantly growing band of brothers who are recognizing its benefits and applying the Monte Carlo approach. It is a specifically suited tool for exploring uncertain events to estimate probabilities. The expanding field of advocates are found in the Mathematics, Physical Sciences, Computational, Engineering, Business, Financial, and Retirement Planning communities. From its limited World War II era introduction, it is now a ubiquitous tool.
In an uncertain environment, having some formal procedure to estimate the success odds of any project and its options is of paramount importance.
As behavioral researchers Belsky and Gilovich remarked: “Odds are, you don’t know what the odds are”. In some sense, investing is a form of gambling. Award winning economist Paul Samuelson cautioned that “It is not easy to get rich in Las Vegas, at Churchill Downs, or at the local Merrill Lynch office”. However, investing is not a Zero-Sum game. Odds can be tilted to favor the patient, prudent, and informed player.
The recently discovered superior Monte Carlo simulator is from Flexible Retirement Planner. Please consider exploiting this especially useful aid to the retirement decision process:
http://www.flexibleretirementplanner.com/wp/or more directly to the simulator itself:
http://www.flexibleretirementplanner.com/wp/planner-launch-page/It is very fast, very flexible, and very worth a visit. This particular Monte Carlo code was written by an experienced, practical, retirement specialist. The calculator’s organization clearly demonstrates the benefits of his hands-on experience.
Monte Carlo analyses are the only investment tool that yields a reasonable estimate of the odds for a successful retirement. It certainly is not perfect, but it is far better than a crystal ball. By using it to explore various retirement and investment options, a candidate retiree can adjust his plans to improve his performance.
Understand that Monte Carlo codes never guarantee 100 % accuracy. That’s impossible in an uncertain world full of unknowable Black Swan happenings.
Many industry specialists suggest that retirement be delayed until Monte Carlo simulations forecast a 95 % success likelihood. That means that there is a 5 % possibility of portfolio bankruptcy. There will always be residual risk in retirement. A parametric Monte Carlo analyses helps a candidate retiree to identify and to minimize that risk, not entirely eliminate it.
In some instances, the stock market will turn sour shortly into retirement. That is unfortunate but not fatal. Those retiring just before 2008 suffered that nightmare. No mechanical tool, no soothsayer could have forecasted that scenario. Don’t indiscriminately scapegoat the analytical tool for the Black Swan physical happening.
Please take advantage of this outstanding resource. It will be both a learning experience and an opportunity to assess your portfolio’s survival odds. Also, I suggest you do a few “what-if” exploratory cases to examine potential pitfalls and improvements. The referenced code makes that an easy chore.
Good luck guys. Some folks might even perceive running these codes as fun.
Anyway, I have fun making the Monte Carlo case. I shall now go quietly and happily into the night.
Best Regards.
Comments
You and I have had this conversation before. Monte Carlo simulation is certainly advantageous, and had it been available in the timeframe which was crucial for us I would certainly have employed it. As it was, I did utilize some of the best planning tools which were then available, from companies such as Vanguard and TR Price. Using those models, it appeared that everything was going to be just fine.
However, those models did not have a variable input for "black swans", aka Murphy. Having worked for many years under an engineer who did not share my perspective, and who evidently was incapable of learning from the experiences of reality, I regard any attempt to model or forecast any complex outcome, without examining the potential for Mr. Murphy's intercession, fatally inadequate. I would reference the construction of the San Francisco Bay Bridge, or Boeing's problems with electrical systems as current examples of engineering failure to anticipate Mr. Murphy's pernicious but predictable effects on complex systems.
Using our financial spreadsheet, I constructed a series of tables which allowed the effects of compounding over a 40 year span to be considered. Additional tables dealt with potential income projection from investments, cash reserves, Social Security and pension sources. All of these included inputs which allowed for various rates of annual income, inflation, allowed variable income and drawdown factors for both equity and bond investments, and most important, allowed for the introduction of a Murphy event, of varying devastation, at a chosen time. And that event did in fact occur, to no great surprise on my part.
To suggest that merely because someone disagrees with your conviction that Monte Carlo is the ultimate planning tool they are "predisposed against statistical analyses" or regard such tools "as far too mathematical, too exotic, too sophisticated" is, frankly, insulting. It is a decent and useful tool, but hardly deserves the degree of faith with which you endow it. 95% is very nice, but it definitely is not reality.
From all of this you may intuit that I am not an optimist. You would be correct.
Regards- OJ
Regards,
Ted
http://www.caniretireyet.com/the-3-best-free-retirement-calculators/
The use of Java doesn't bother me - a decent browser will block all applets unless you tell it you're willing to trust the site (as noted in the article cited by Ted). Sure there can be bugs in the applet, but that's different from suspecting the code of being designed to do "bad things".
On the other hand, I remain bothered by the general lack of specific details on the data set used, and how that data is used (e.g. rolling five year periods, sequential periods, monthly, etc.) More generally, to the extent that samples are selected at random, there's an implicit assumption that data points are unrelated. In the real word however, various factors are not completely independent - given persistence of performance, increasing correlation of asset classes in down markets, etc. Maybe the simulations are more sophisticated than I give them credit for, but I can't tell that.
I like tools that deal with the "opposite" of a black swan event - known major expenses at a particular point in the future. For example, I expect to buy another car in about a decade - maybe a Tesla if there are enough recharging stations by then (or a Prius, depending on how the market goes)
Another nice MC simulator is T Rowe Price's FuturePath, though it seems you need to have an account with them to create a login which is needed to use the tool.
These tools are good for "what ifs". For people like OJ, they're not going to give you warm fuzzy feelings, but they can help with planning, learning what strategies generally tend to work better than others.
I want to thank the MFO members who took a timeout to read about my enthusiasm for the Flexible Retirement Planner. I have been a Monte Carlo cheerleader for decades ever since my military days when participating in war games planning exercises.
An especial thank you to those MFO members who contributed additional Links and opinions. Good decision making demands open and fair debate that exposes all options.
Monte Carlo analysis is a relatively modern statistical methods addition. It is rooted in the uncertainties associated with the development of the atomic bomb in the 1940s. John von Neumann and Stanislaw Ulam are usually credited with pioneering formulations.
Monte Carlo analysis has penetrated many military, industrial, and scientific disciplines. It actually received financial planning attention rather late in its brief history. Today, anyone seeking this planning tool for that purpose has the luxury of many excellent choices. It’s good to have choices.
I retired in 1994. In the early 1990s I searched without success for such a tool. Not finding one, I programmed my own retirement Monte Carlo code. Somehow (I don’t recall the source), I became aware that Nobel Laureate Bill Sharpe was also working in this arena. I contacted him for advice and he responded with first-aid suggestions several times.
A little later, Professor Sharpe became a founding father of Financial Engines. Financial Engines is powered by an excellent Monte Carlo code. It went public in 1996. Since that introduction, it has developed working relationships with the big three mutual fund houses (Fidelity, Vanguard, T. Rowe Price) as well as powerhouse financial institutions like JP Morgan and Northern Trust. The tool has been extensively tested and is constantly being improved in terms of options and basic modeling.
Be alert that these codes use different ways to estimate future returns. Of course, they all use a random selection approach. Some programs use actual returns from the markets and randomly select from that database. Others use statistical mean, standard deviation, and correlation coefficient models when drawing from the returns grab-bag. The referenced Flexible Retirement Planner deploys this approach, but it allows its users several alternate options.
In my code, I too chose the statistical model method. Even in the 1990s, I was a little skeptical and unhappy with the model since it did not include the shocking outlier years (before Taleb dubbed these events Black Swans). The beauty of Monte Carlo modeling is that these outliers can be easily incorporated into the analysis with the addition of a few subroutines. I made the necessary changes using past “Black Days” data to guide the statistics and including some multiplier factors to exaggerate the impact.
The inclusion of these disastrous days did degrade portfolio survival probabilities, but only at the margins. Given the uncertainty of the events, their infrequency, and the ad hoc way I modeled these rare events, I finally discounted that segment of the analysis in my retirement decision. I interpreted the outlier outcomes as noise.
Monte Carlo codes as part of the retirement planning toolbox have been accessible since 1996. Monte Carlo procedures were initially challenged by advisors who were a little short on a mathematical and/or scientific background. Most of these reluctant advisors were finally converted given the power and success of the tool. I will not name names, but many respected financial institutions were participants in this eureka story. Monte Carlo based tools are now universally applied in the retirement planning industry.
I encourage you to join the parade. In the end, you might not embrace the output, but it can do no harm because the final decision is always yours to make. It’s good to be King.
Best Wishes.