by Jack Paul, CLU, ChFC, CASL, FSA
Jack Paul, CLU, ChFC, CASL, FSA, is president of Jack P Paul Actuary, LLC (www.JackPaulCASL.com), a consulting firm for financial planners. He is a fellow of the Society of Actuaries and member of the American Academy of Actuaries. Contact him at Jack@JackPaulCASL.com.
Monte Carlo testing, also referred to as stochastic testing, is often performed for clients of financial planners as a way to calculate the probability that the clients will be able to meet their financial goals, as well as to validate spending and investment strategies to support those goals. The use of this sophisticated technique is fairly new, being made possible by the advances in computing power and by certain software packages that can be used/leased/purchased to perform the testing. For clients at or nearing retirement, there are some very important considerations that are just starting to be incorporated into this type of Monte Carlo testing, and which have a dramatic impact on the calculation of those probabilities. This article will describe these considerations. Although a client will be assumed to be a single male in this article, the considerations equally apply to couples.
Briefly, the type of Monte Carlo testing used by financial planners involves calculating hypothetical asset rates of return over the future life of the client (based on the characteristics of his asset portfolio). The client will have an anticipated spending strategy over his lifetime to support his financial goals. The asset returns are combined with his anticipated spending, income, and other factors to determine if he will be able to successfully meet the goals for the given asset rates of return. This process is repeated many (hundreds or thousands) of times to determine the probability that the client will meet his financial goals. If the probability is too low, the asset portfolio or the spending strategy can be modified to increase the probability of success.
One important consideration that has recently become introduced into this type of Monte Carlo testing is the variation of the assumed time of death of the client. Often, testing is done assuming the client will live to a fixed age, such as (for a 65 year old) age 85, 90, or 95. Often the age picked is the assumed life expectancy of the client. Of course, death can occur at any age.
A second consideration is the variation of the timing and amount of the client's future long-term care costs. Often, long-term care is modeled as a single event, such as a two-year stay in a nursing home at age 80. In reality, the need for long-term care can occur at any time (although the most likely time is after age 75), and costs can vary over the client's lifetime, from zero to well over a million dollars. And a third consideration is the variation of the timing and amount of the client's future prescription drug costs. Often these costs are assumed to be the current costs increased with inflation, but as a client's health changes over time, the additional drugs that will be needed will cause these costs to increase. These costs can vary from very little to more than a $500,000 over the client's lifetime.
These considerations are now beginning to be addressed by incorporating the client's potential long-term care and prescription drug timing and costs into the testing, as well as by varying the time of death (by the use of mortality rates). This expands the testing from what I'll call "Monte Carlo asset testing" into "comprehensive Monte Carlo testing."
Comprehensive Monte Carlo Testing
An important feature of comprehensive Monte Carlo testing is that it is being customized to the clients' unique morbidity and mortality profiles. Screener questionnaires are filled out by the client, with the help of the financial planner, in order to produce accurate probabilities of long-term care usage and prescription drug usage, as well as the probabilities of living to various ages.
This comprehensive Monte Carlo testing incorporates the client's spending, asset portfolio and investment strategies to go along with the long-term care and prescription drug potential costs, as well as the potential mortality of the client to give the client a comprehensive picture of major retirement risks. The risks all are combined into one useful, meaningful measure-the probability that the client will meet his goals, including the all-important goal of not outliving his assets.
Comprehensive Monte Carlo testing is a very flexible and powerful tool. For example, the financial planner has the option to work with a client to produce a customized level of long-term care (should the need arise) as input into the testing. Would the client want a private room in a nursing home? If the client wants to remain at home and never go into an assisted living facility or nursing home but instead wants a nurse at home 24 hours a day, how does that desire affect the client's ability to meet his goals? What would a more modest level of care look like?
This testing can open the door for a financial planner as it is possible to examine different insurance strategies, such as long-term care insurance policies and/or riders, longevity annuities, prescription drug plans (including Medicare Part D plans) and other products to see the effects on the client's goals, and to perhaps result in a sale beneficial to both the client and the planner.
To arrive at an acceptable outcome, the financial planner works with the client to run iterations of the testing, examining changes in investment, spending, insurance and other strategies to produce acceptable results for the clients.
Illustrating the Differences
Here is a brief example to show the difference between Monte Carlo asset testing and the comprehensive Monte Carlo testing as described in this article. The differences are illustrated using a 65-year-old single male. The differences displayed are in the assumptions used and in the results.
Time of death:
Monte Carlo asset testing: Age 90
Comprehensive Monte Carlo testing: Varies based on a mortality assessment
Long-term care costs:
Monte Carlo asset testing: A two-year stay in a nursing home starting at age 80Comprehensive Monte Carlo testing: Varies based on a morbidity assessment
Prescription drug costs:
Monte Carlo asset testing: Current drug use assumed to continue throughout life.
Comprehensive Monte Carlo testing: Drug use could change based on a morbidity assessment.
Results, expressed as the probability that the client will not outlive his assets:
Monte Carlo asset testing: 58%
Comprehensive Monte Carlo testing: 81%
What accounts for the difference in results? Below are the three main reasons:
First, the assumption in the Monte Carlo asset testing as to the time of death (at age 90) is fixed, ignoring the fact that death can occur at any time. The 65-year-old male portrayed in this example has only a 37 percent chance of reaching age 90. The longer-than-average assumed lifetime requires many years of spending, overstating the chance that the client will run out of money while alive.
Second, the assumed event of a two-year nursing home stay at age 80 is relatively costly compared to all the different possible long-term care events that could occur. In fact, lower long-term care costs are incurred 87 percent of the time. Again, this overstatement of costs causes an understatement of the chances that the client's assets will last for his lifetime.
Third, these two overstatements of costs are offset by the understatement of prescription drug costs. The Monte Carlo asset testing assumes that the prescription drug costs only increase by inflation. In reality, the client can develop new chronic conditions over his life, possibly increasing the use and cost of prescription drugs over and above inflation.
Why Are These Considerations Important?
These two probabilities-81 percent as opposed to 58 percent-are quite different for such a critical computation. Using the additional considerations described in this article gives crucial insight into what is often the most important issue in the client's mind-whether his assets will last for the rest of his life. This additional insight is crucial. If the client's chance of success is understated, there could be an unnecessary cutback of the client's lifestyle. And if the client's chance of success is overstated, that could lead to a false sense of security and could result in a client running out of money because of the overstatement. Therefore, it would benefit financial planners to consider incorporating comprehensive Monte Carlo testing into their practices.
Article copyright 2010 by Jack P Paul Actuary, LLC