Monte Carlo simulations have become the dominant method for conducting financial planning analyses for clients and are a feature of most comprehensive financial planning software programs. By distilling hundreds of pieces of information into a single number that purports to show the percentage chance that a portfolio will not be depleted over the course of a client’s life, advisors often use this data point as the centerpiece when they present a financial plan. However, a Monte Carlo simulation entails major statistical and philosophical nuances, many of which might be underappreciated by advisors and their clients.
One key nuance to the use of Monte Carlo simulations is whether they are being used as part of a one-time plan versus an ongoing planning process. For example, a Monte Carlo simulation resulting in a 90% probability of success will mean very different things depending on whether a client will take fixed portfolio withdrawals throughout retirement based on the initial probability of success or whether they plan to run additional simulations over time and are willing to adjust their spending based on market performance. For the former client, because a 90% probability of success means that there is a 10% chance they will deplete their portfolio (though the magnitude of the failure is unknown), they might choose to aim for an even higher probability of success to decrease the likelihood that they will run out of money in retirement. But for the latter client, to suggest they have a 10% chance of depleting their portfolio is overstating the risk, as they are willing to adjust their spending in response to future simulations that show a reduced probability of success.
An alternative way to use Monte Carlo simulations for clients who are willing to be flexible with their spending is to consider how spending would change when using a fixed probability of success. For instance, Monte Carlo simulations show that, for any selected fixed probability of success, the maximum and minimum annual spending for a client during the course of their lifetime is remarkably similar. While initial spending levels will be different depending on the target probability of success (as a higher selected probability of success will call for a reduced initial spending amount), adjusted spending levels will track each other closely no matter the initial probability of success chosen. What is different is that those who use a higher constant probability of success will likely have a larger portfolio balance at their death than do clients who choose a lower probability of success at the start of retirement.
This suggests that, in contrast to the view that probability-of-success levels are indicative of the risk of depleting a portfolio, the probability-of-success level used when adjustment is planned for in advance is essentially akin to putting your thumb on the scale to slightly favor either maintaining current income (lower probability of success) or preserving estate balance (higher probability of success). In other words, if an advisor is going to use Monte Carlo on an ongoing basis, then the probability of success threshold targeted is more akin to a slider that adjusts the degree of preference for current income or legacy rather than a meaningful measure of the likelihood of depleting a portfolio.
Ultimately, the key point is that because the results of Monte Carlo simulations contain a significant amount of nuance, particularly if being utilized as part of an ongoing planning relationship, advisors can consider using them as an internal analytical tool but communicating the results through the use of risk-based guardrails or as a tradeoff between current income or legacy interests to help clients better understand what the results actually mean for their financial plan!
Leave a Reply