When building your financial plan, you can choose to build a model that illustrates the range of what might probably happen, or you could build a model to illustrate what you believe will happen "for sure."
So what is the most appropriate way to build your lifetime planning model in MaxiFi?
MaxiFi Planner provides two complementary approaches to planning:
- A deterministic plan intended to be used with safe assumptions answers the question: How much will I have available to spend "for sure" or with reasonable certainty?
- A statistical (aka, probabilistic or stochastic) plan that reveals a full range of possible outcomes based on your asset allocation and answers the question: What are the probable ranges and trajectories of living standards I might have over my lifetime given my spending behavior and my asset allocation?
The first approach (deterministic) is what economists refer to as a certainty equivalent plan and is what you build when you create the Base Plan or non-Monte Carlo Base Plan report. The second approach (statistical) creates the Monte Carlo Risk Analysis report.
Because the first approach is deterministic and assumes that you will always get the rate of return that you enter and always spend and save the recommended amounts, it is appropriate and prudent to enter return assumptions you feel you can get for sure year over year, not a historical average. If you assume return rates you are not relatively certain you can reliably get each and every year with predictable certainty, you will introduce risk to your plan. You will no longer have a safe plan. Using return rates you are not relatively certain about year over year will mean you are working at odds with the purpose of a deterministic approach. The deterministic approach establishes a floor or base profile, a "safety-first plan" and creates a report that you can feel confident about.
The statistical approach draws returns at random based on your investment and spending strategies—how you intend to invest your assets through time and how cautiously or aggressively you intend to spend. You can invest aggressively or conservatively. Same for your spending behavior. The more risk you undertake in investing and the more you spend in the short run, the greater the chances of a significantly lower living standard down the road. Our Monte Carlo living standard analysis shows you the range of outcomes you may experience. You will see these outcomes in the full context of the statistical upside and downside in your plan. This probability-based planning approach has an important function, but its purposes are different from the safety-first approach of the deterministic plan.
You are Not an Insurance Company
Consider how an insurance company thinks about how long you might live when they underwrite your term insurance policy. Because they can pool their risk across many customers, insurance companies can work the actuarial averages to their advantage to make sure they are price competitive. In this case, the statistical probability that you will live to 83.5 is highly relevant to their analysis and it may be the age they use. That age is appropriate to their modeling.
But when you create a lifetime financial plan, you have to acknowledge that you are not an insurance company that can pool longevity risk across multiple lives. You have just one life to live and therefore must assume the worst, which from a planning perspective is that you could live to 100. Of course you probably will not live to 100—you will probably die at age 83.5 as the actuary tells you—but, again, you are not an insurance company and you can't plan for probabilities because you are a pool of one and must plan for the worst. The best approach to planning is to start with a deterministic plan that is safe (does not hide the risk from you), and next, if you want a fuller picture, you can create a statistical plan that reveals the full range of upside and downside probabilities in your plan based on the asset allocation you plan to use.
These observations about how being a pool of one is different from the way an insurance company approaches risk reveal the fallacy of the "break-even analysis" that sometimes arises when people think about when to take Social Security. The break-even analysis is conceptually flawed for the purpose of financial planning because it treats the problem as if you can play the averages. It ignores longevity risk from the point of view of the singular life. But when Social Security is viewed as longevity insurance, we realize that we must plan for the possible, not the probable, simply because we can't count on dying on time. Indeed, nearly every kind of insurance is purchased because of what could possibly happen, not because of what will probably happen.
But longevity is not the only assumption we must make in order to build our model. We must also assume something about our real rate of return on retirement and non-retirement assets.
Here again we are faced with a choice:
- Indicate what we think we will probably get in returns (for example, the historical average) over the life of the plan or
- Indicate a safe return assumption.
Using a historical average in the stock market or average for our particular asset allocation in a deterministic plan is analogous to using an actuarial assumption about when we will die. Using a historical average is statistically sound and even appropriate for some kinds of planning, but it's not appropriate to a deterministic financial plan. Why?
Why we should use safe returns in a deterministic plan
An individual deterministic financial plan creates a plan that represents itself as precisely meeting all expectations in the future. It assumes you will get for certain the return that you stipulate each year and assumes you will die at the age of your life expectancy. But it does not account for any variability in the future, such as living beyond your life expectancy or achieving returns below the stipulated rates. So for the deterministic approach to be useful in planning, it is essential to establish a “for sure” scenario that represents a future that is certain to be attained. This for sure model is constructed by using safe assumptions that can live up to this billing and the results serve as a baseline or floor for retirement spending. The higher you set this nominal return assumption the higher will be your "safe spending floor." If you set it too high, you will not have a safe spending floor, you will instead have a "maybe spending floor" or "risky spending floor."
Why is this important? The planning model guides your actions: it guides you in your decision to work longer or not, buy this house or that house, take this job or that job, save and spend this much or that much. If your plan was just something you were curious about and not a guide to action, then any assumption you make might satisfy some statistical curiosity. But because you are making real life decisions every year that you can't take back, and because the deterministic plan represents its singular plan outcome as a nearly certain plan instead of as a range of 500 possible outcomes like in the statistical plan, it's important that this single outcome be a reliable outcome guiding you with cautious saving and spending recommendations, leaving you with annual, incremental upside potential rather than possibly putting you in a financial hole.
Why then does conventional planning software ask you to enter your expected, average rate of return as opposed to safe rate of return? Why is MaxiFi different in this regard? With conventional planning, the approach is not designed to spend all of your money down to zero dollars. Conventional planning cannot discover your built in (endogenous) spending level and so asks you to estimate or guess at your annual spending level and then tries to determine the statistical chances of not running out of money by leaving unspent money on the table representing the slack needed to make your model safe. In this sense, conventional planning is sloppy. It leaves a lot of slack in the plan as a safety net and presents its conclusions to the user as "X% chance of not running out of money." All questions you ask of conventional plan are answered this way: "How do you feel about having X percent chance of not running out of money?" MaxiFi planning, on the other hand, discovers your annual spending level and begins with a deterministic "for sure" approach based on safe assumptions and spends all the money that you don't set aside in reserve funds or bequests or housing.
With MaxiFi, the plan is designed to never run out of money. Because it discovers (rather than asks you to enter) your annual spending level, the deterministic plan is a for sure plan because it is not based on a user's guess about how much is needed to spend each year. This never-run-out-of-money design, is built on a platform of safe assumptions about longevity and rates of return (and a few other decisions as well in the inputs). Note that MaxiFi does not try to combine deterministic and statistical approaches in one hybrid plan. Conventional planning asks you to set annual spending level as a deterministic, user-entered (exogenous) variable, but then in the same model, runs a statistical analysis. MaxiFI discovers your built in (endogenous) annual spending level and allows you to run a statistical risk analysis separately.
What is a safe assumption about rate of return? The answer is admittedly subjective—people have different levels of risk aversion—but a good place to start is a .5% or 1% real return and with recent events in the economy related to COVID-19, MaxiFi now defaults to the cautious suggestion that you use a 0% real rate of return—that is, a nominal rate of return equal to inflation.
Most users understand that assuming life to age 100 is a safe assumption. But they sometimes balk at assuming some analogous safe assumption about nominal rates of return. Perhaps this is because the financial press and planning industry puts so much emphasis on average rates of return over long periods of time and conventional planning bundles all the risk in left over unspent assets at the end of the plan. Or perhaps we just don't want to concede that we might not be "above average."
Whatever the case, building a plan with a safe assumption about longevity and nominal returns and cautious spending assumptions is the appropriate way to build your deterministic base plan.
If you use historical average rates of return, you must look at your model and add an asterisk indicating: "this plan assumes I get the the historical average return which I might not get." Instead you want to view your plan and say to yourself, "this is a plan built on safe, cautious assumptions about longevity and rate of return and it's safe and reasonable to assume that I will do at least this well." If you use rates of return based on historical statistics, to be consistent you should also use your probable age of death, let's say 83. But then, if you live until age 84, your plan, as designed, will have failed. An analogous fail will happen if you don't get the historical average return you expect, and there's a good chance you will not. The Monte Carlo or statistical approach is designed to show and illustrate such consequence which is different from the deterministic approach where using historical average hides the risk and is not explored and shown in the deterministic reports. The deterministic plan bills itself as a really safe plan; therefore, you should use really safe assumptions to remain consistent with this purpose and live up to this billing.
Are you giving up on or walking away from any upside in your plan?
In the Base Plan, the deterministic plan, you may well be low-balling your plan by hedging your bet using these safe assumptions, but it's some consolation to keep in mind that if safe assumptions about return and spending behavior cause you to under spend your model because your actual returns going forward exceed your "safe assumption" and "cautious spending behavior" you will not forever miss out on that upside. Instead, as you rebuild your model at the start of each new year, you will incrementally include your unspent gains from last year's good returns into each new year's model. All things being equal, you would thus observe your annual spending rise in real terms each year as you shovel forward unspent gains in regular and retirement assets each year should they come to pass. The point is, the smart approach is to realize this upside incrementally, year over year, beginning each new year when you rebuild the model. You will realize your good fortune, should it come along each year, only as it happens instead of baking it into the model as a supposedly reliable, safe assumption about the future at the outset.
Probabilities are Still Relevant
Statistical probabilities, however, are still relevant. When building you deterministic base profile, you want to use safe assumptions because it is a deterministic model. With Monte Carlo Risk Analysis you will draw on statistical probabilities and see a range of possible outcomes to your plan. It's important to separate these two planning purposes, deterministic and statistical. Trying to combine them by using historical averages in a deterministic approach is a source of confusion and the results can misrepresent the "for sure" character of the deterministic model causing you to think that the average must be your destiny because, well, it's the average. When you use the Monte Carlo analysis, you use historical returns and risk and can see the results of your historical average rate of return, but you see this average trajectory in the context of a full set of 500 statistical possibilities. Claiming that this Monte Carlo analysis was somehow more accurate might be like claiming that one of the 500 possible annual spending trajectories it presents is more accurate than all the other 499 or than the flat trajectory of the deterministic plan. Remember, you don't get to pick which of the 500 trajectories outcomes might be yours. It could be any one of them with various degrees of probability. A Monte Carlo model, though it gives a lot of new and useful information, is not more accurate than a deterministic model. It reveals the range of what is possible given your asset allocation and spending behavior. But a deterministic plan when set up with safe, or nearly certain assumptions, shows you your safe annual spending allowance, an annual spending cap that is actionable today. The two approaches complement each other.
When you assume the historical average in the deterministic approach and bake it in to the base plan, you see the result as a single answer, possibly masking its risk, and you risk convincing yourself, wrongly, that the historical average is showing you your safe, reliable reality. The average is not your certain reality, even over long periods. Over long periods of time many plans will really do much better and many will really do much worse than average. As the Monte Carlo analysis reveals, there are reasonable chances that your living standard is much lower (or higher) than this average. Again, if you were not making decisions each year that you can't take back, all of this would be academic, it would not matter. But you are deciding how much to spend and save each year and making other important decisions like when to take Social Security or how long to keep working. Thus, you must be cautious, not "play the averages." Seeing a display of the statistical probabilities that are a result of your asset allocation can give you a picture of how much risk is in our plan.
In Summary
The purpose of a deterministic approach is to show you annual saving, withdrawal, and spending levels based on cautious spending and safe return assumptions. When you are guided by these safe returns and spend "as if" you get these safe returns for sure, you can also keep in mind that you will rebuild your model each year to reflect actual annual gains or losses in the market along with new economic assumptions. It's important to keep this deterministic approach separate from the purpose of discovering what your spending would be if you earned the market's historical average. If you are interested in historical average returns, or interested in exploring through statistical analysis the risk in your asset allocation, you should use the Monte Carlo risk analysis where these historical returns are presented in the context of the full range of possibilities thus avoiding the fallacy of thinking you will always get the average, which is a risky assumption when you are making saving and spending decisions you can't take back.
See also: Inflation and Rate of Return Data
See also: Spending Behavior in Monte Carlo