Just where does the fear of outliving our money come from? Part II.
The foundation to answer where does fear of outliving money come from was laid in yesterdays’ post (Part I).
Why is all of this important? The fear of outliving your money also (there’s another reason listed below) comes from behaviors of the present self not considering the impact of present actions on your future self.
Chile’s pension system is in dispute because people are not saving enough (“The perils of not saving” in The Economist). Sound familiar? People are blaming the system instead of their own actions – under saving. The corollary is over spending based on resources actually available.#
What’s the problem? Saving too little while working with the expectation of spending too much while retired. Folks – it doesn’t work that way!
People (us), pension plans, insurance companies, mutual funds, in other words everybody and every entity, can not control the economy or markets within which we all live. Those are the sum of all our actions and we can’t control the actions of everybody else on the planet living within the economic system.
Everyone is a participant as a subset of the whole. Everyone invests in the same markets.
How can we approach these fears rationally?
You control your own actions and face the consequences of those actions. The main objective of the research in yesterday’s post is to develop a more robust model than currently used in the profession for connecting today’s spending while retired to the future year potential capital range and thus spending range at those future ages.
However … the current practice for calculating or simulating prudent income is not much different than long division (see graphic at beginning) when viewed as having a solution (how much can you spend today), the derivation of that solution, and a remainder value. Basically the derivation and remainder values of today’s calculations/simulations are interpreted as possible future states. The research on yesterday’s post shows “possible future states” (remainder values) is not a valid interpretation of the data because the subsequent year’s solutions would be different; because the time frame for those solutions are different due to aging and period life table statistics, not fixed as viewed today as a result of calculation methodology.
What does the Monte Carlo (stochastic) graph look like when ALL points on the graph are the solutions, and NOT the derivations? You get a graph that looks like the one on the right.
What’s the difference between the two figures? The bottom figure is made up of ONLY solutions, while the top graph is only a single solution (the rest of the graph lines are derivations for that single solution). The top graph provides a wide range of the future, while the bottom graph narrows the future down.
What the research project found* was that making adjustments to spending resulted in capital available for future spending even into old age, unless 1) either a catastrophic spending need occurs, or 2) there’s an unwillingness to change spending when signals suggest it would be wise to do so. Of course, if all goes well then the fear of running out is unfounded – except, then again, the appearance of having a large excess may also lead to excessive spending relative to what is needed for your future self (think lottery winners who are worse off than before because they spent it too fast – overspending again).
The theme? Today’s calculations are NOT a model that signals to your present self how much you can spend from a portfolio, and retain funds for all your future selves. Essentially, the recalculation each year computes how much money you need for the rest of your life – with “the rest of your life” coming from percentiles of period life tables. In other words, “Personal Mortality Credits” from your own money that you are free to bequest or spend. Today’s calculations (simulations) only suggest what is prudent today, not what effect today would have on yourself at a future age (yes, calculations/simulations provide a range of possibilities, but not an age-derived solution).
Programming and software capabilities exist today to develop more robust models to address this. The body of thought today has not taken the next step from today’s single period calculations, that provide just today’s answer about spending, and don’t provide insight into future year spending other than through a wide range of future possibilities (we have to wait until we get there attitude).
In other words, today’s retirement perspective arrives at a solution, but then tries to read into the future through interpretation of the derivation and remainder values of that single period solution; like trying to read tea leaves at the bottom of the cup.
The research project in yesterday’s post demonstrates that the range of future possibilities can be narrowed by limiting cash flow estimation errors over time, as illustrated in the above graph, by calculating future possible spending for each potential future year and stringing those calculations together (second graph). Each age has its’ own separate solution which need to be strung together as solutions; not a single solution with its’ derivation and remainder values (as illustrated in the first graph) which really don’t suggest much about future spending based on future ages and expected longevity tied to those future ages (time remaining slowly decreases and this affects cash flow from those calculations).
* “Floors and ceilings”
The research project observed that when spending caps (ceiling) on the upside were put in place, then the portfolio balance tended to grow much more , due to the spending rule, relative to when there were no spending caps. Also, when a spending floor was put in place, then the portfolio balance tended to experience “failures,” or money ran out at some point in time due to the inflexible spending rule in place#. Single period simulations have both a cap (ceiling) and a floor in place because the single spending amount leads to either balance growth during good simulations (good markets), or running out of money during poor simulations (poor markets). In other words, a set spending amount with or without inflation adjustments, is essentially both a spending floor and ceiling. This leads to a “fan shaped” simulation result, as commonly seen in the first figure above, rather than the teardrop shape seen in the second figure above.
So where does the fear of running out of money come from?
Moral of the story for the answer: When I was studying for my CFP certification, the focus was on deterministic calculations. Since then stochastic process has taken hold to incorporate randomness (illustrated in the top figure); but, still over a fixed, single period of time, and a sole solution derived. A simulation over a fixed period may have a percentage of simulations that fail – and this is my argument for some of the fear of running out of money; in other words, through such an approach, the profession suggests to retirees that they might based on the methodology and interpretation of results. The research referred to in the past two blog posts determined that the possibility of running out of money came from an inflexible minimum requirement.
Instead, the profession should adopt a more positive model explaining how spending may be managed so as not to run out of money, and software developed to support such a model. Of course, when the future arrives, new data and facts should be incorporated as an update and review. The profession’s dialogue with retirees’ would thus reduce or eliminate the concept of running out of money!
Software processing speeds, and programming capabilities are now at a more advanced stage, and a tipping point has been reached, where multiple stochastic solutions, each for different ages, may be calculated to form a model that does not end until the period life table ages end. The model approach illustrates the possible range of income available at those later ages, thus reducing the fear of outliving the money. The model approach also more directly illustrates how spending more at younger ages may affect the ability to spend at older ages (as well as what bequests may be at various ages as well); baring catastrophic spending anywhere along the age spectrum.
So what should advisers, academics and clients do in the meantime pending more robust software programming?
- Perform an annual recalculation of prudent retirement spending from the portfolio using updated data for both the portfolio and the life tables (bottom graphic above).
- Refrain from inferring from solution derivations and remaining values what might happen in the future (top graphic above).
This research project suggests that modeling (multi-casting) can be refined to do a better job evaluating future expectations than the current single-cast simulations do. Fear of failure should be, and can be, managed better.
PPS. When you read about retirement income from a portfolio, you get the feeling that ALL your income is sourced from your portfolio.
- Those who are already retired also tend to have Social Security and/or a pension as well. Thus, for example, if half of your total retirement income comes from Social Security, then the other half comes from your portfolio to equal your total. Therefore, if market misbehavior pulls your portfolio down by 25%, then the effect on your total income would be 12.5% (half of the portfolio because the other half is not market based). How you allocated and diversify (they’re two different things) does affect the fluctuation of portfolio value so the approach for this is important (evidence and Dimensional). If 12.5% is too much of a portfolio decline, then this is a good signal to have less exposure to stock markets so the possible portfolio value decline is reduced because of that lesser exposure to decline. See point 2 below too.
- Sequence of returns risk. Most people make the mistake of setting the spending level from a portfolio based on what I call “the economist’s view,” where taking the most possible at all times is the goal. However, because the markets misbehave on occasion and go down in value, this means adjusting income down when markets go down, and back up when markets go back up. Even on an annual level for these adjustments, that is not how most people’s budgets work. This is following the “Ocean Wave,” or the ups and downs. Instead, if you calculated regular spending on an adjusted portfolio value that is based on the possibility of portfolio value decline, then the portfolio value can go up and down all it wants without affecting your spending, at least until it goes below the value you calculated, and then it could go below that even just a little bit too, with maybe just a small spending adjustment. However, most of the time markets tend to rise, so there may be an excess of portfolio value above your calculated amount that could be available for “discretionary spending.” So why base spending on a reduced portfolio value instead of simply reducing spending in the first place? Because … basing it on a lower portfolio value allows you to compare your present portfolio value, as it goes up and down with market forces, to the present “trough” portfolio value … this way you can “see” when a small possible additional spending reduction may be possible … or when additional “discretionary” funds may be available as well. Without the portfolio value as a reference, it is hard to measure and monitor the spending threshold.
- Point #2 is also how to address the fear of “sequence risk of returns.” Many view this fear as temporary and it exposure to the risk goes away with time, mainly because the portfolio balance MAY grow sufficiently to the point that you are underspending relative to what the total portfolio value suggest you could spend with that balance. In other words, spending is set low relative to the balance from the beginning and kept in check through a rule of thumb inflation adjustment. My experience with retirees at all ages shows me that ALL retirees fear the risk of their portfolio going down “this year.” Point #2 addresses that fear.
- “Personal Mortality Credits.” By using a long calculation period, the result is that a low percentage of people may outlive that long retirement period elected. When such a long period is applied to everybody (all clients), this is akin to “punishing” (restricting income for) the many, in order to have income later for the few who do reach that much older age relative to the period life table. A model approach that uses rolling time periods from life tables encourages people to spend money during the years they’re more likely to be alive to spend it; with slow adjustments over time to accommodate the life table increase in likely ages the retiree may reach, combined with the tendency to reduce spending with age (until health tends to increase expenses again for some – not all – retirees … this suggests insurances (covering a potential risk) should be the solution, rather than requiring everybody to reserve for expenses they may never incur; which is the result of current practices in the profesion).
#Inflexibility when it comes to spending, especially with a minimum spending requirement that is fixed (a spending floor), resulted in money running out before the end of the time period(s), i.e., failures. This finding has significant implications for pensions and annuities which have just this type of spending requirement – fixed and inflexible. Pensions and annuities may not be as safe as prior fixed period calculations or simulations because of these rigid spending requirements.