Many people think of retirement as passive where they don’t need to do an annual update anymore. So what does an annual checkup do?

I suggest that annual reviews are critical during retirement. Yes, there is uncertainty about the markets going forward (I call this the **Probability of the Portfolio**). Yes, there is uncertainty about how long you might live (I call this **Probability of the Person**). Both probabilities need to be considered together and integrated into how you determine your retirement income amount each year. I’ll discuss both in general below, starting with longevity.

The purpose of a annual review is to update the facts as of that review date. This includes updating the market returns data that simulations and calculations are based on. This includes updating your portfolio values. This includes updating your spending needs. This includes updating your expected longevity. Why? Because that longevity age provides a better reference than a guess does as to how many years your retirement income may need to last. And, as I’ll illustrate below in the last graph, the number of years you have to your *current* expected longevity age may be fewer than the actual number of years you actually live … because … expected longevity age also slowly increases as you age too.

**Probability of the Person.**

First life expectancy and how it helps establish a good reference point to plan with. Let’s take a couple ages 65 and 60.

Joint ages 65 Male (Green)/60 Female (Blue). Expected longevity either alive is approximately 92. Note ages for the 50% probability line for male or female. Note that expected longevity **by definition** is 50% probability. This means that half of people at a given age outlive the age(s) along the 50% line and half of the cohorts that age do not. Lower probabilities are more conservative, and higher probabilities are less conservative (more on this below).

Joint ages 80 Male (Green)/75 Female (Blue). Expected longevity either alive is about age 93. Note the ages for the 50% line for male or female.

Husband passed away and now a single 88 year old Female (she is sole client – line green). Expected longevity is about 94.

She has now reached age 94 and updated expected longevity is about age 98.

Probability Graphs Source: MoneyGuidePro. All calculations based on Annuity 2000 Mortality Table.

As longevity tables change, because we live longer, this effect is also captured during the review process by using future updated period life tables. For example, the 2000 Annuity Table has recently been updated. Social Security also updates their tables periodically.

Moral of the longevity story: Those who continue to live show, by living, that they *were* *among* those with the higher chance of continuing to live a long life. Our woman above started out with expected longevity of about 88 years old, and may live into her early 100’s, or even older. She didn’t know this when she was young, and she doesn’t know now if she’ll be a centenarian or even a super centenarian. Not until she gets there.

Now, this was presented initially as a couple who progress through life. However I, like many advisers, have many other couples, and singles, at all retirement ages. A generic rule of thumb can’t be applied to all of them because each is unique in their situation and ages. That is the value a planner provides – recognizing your uniqueness and planning accordingly.

Note: a properly managed portfolio most likely will have some value remaining when our female example passes away. Thus, having a will and living trust and other estate plan documents is part of a properly planned retirement plan to transfer remaining assets. Of course, medical issues may require the retiree to spend down their assets (or have health insurance and Long Term Care insurance in place to transfer this spending risk). We don’t know which scenario will play out … so prudent planning addresses both scenarios.

Here’s Social Security’s life expectancy calculator for individuals.

Kitces Nerd’s Eye View has a blog posting on the longevity for a more advanced take on the topic of Longevity Risk. Kitces also has an excellent blog on interpreting probabilities.

Although I present 5 year or more increments in time for the longevity above, the change in longevity is slow relative to change in portfolio values.

**Probability of the Portfolio**

My peer reviewed published research in the *Journal of Financial Planning* (Model built and discussed in issues Dec 2012, March 2012, and Nov 2011) discusses how the two probabilities of dynamically aging and managing a portfolio’s cash flow fit together *in one model*. The bottom annual cash flow graph shows spending money may be sustained (yes more in better markets – 5th and 25th percentile; and less in poorer markets – 75th and 95th percentile) into very old ages if the retiree doesn’t go on a spending binge to deplete their portfolio value at any given time more than their spending need. The green line below is the 50th percentile market results from the simulations.

The spending period in the below graph## is designed for our 65 & 60 year old couple above – and starts out at 22 years for the simulation distribution period length. If you look at the top graph above, 70% of 60 year old females (blue line) outlive age 82 (age 60 + 22), and 40% of males (green line) outlive age 87 (age 65 + 22). That may sound aggressive using just 22 years for them. Now notice below that the cash flows extend into their 90’s and even into their early 100’s if necessary.

22 years wasn’t aggressive after all because, you see, a withdrawal rate is spending a *percentage* of the portfolio value, not *all* of the portfolio value, at any given moment in time (in physics this is called an exponential rate of growth/decay). Therefore, money lasts longer than constant spending suggests if this method is applied for income from a portfolio.# All of the serially connected distributions (cash flows) are calculated at the 10% percentage of simulation failure target level (40% equity/60% bonds).*

The below is just one example and, as I’ve stated above, each retiree and their situation is unique. Structuring a prudent spending plan should be customized and managed based on their uniqueness. There are many possible and prudent spending plans such as spending more during young retired years and cutting back when older; or spending some during retirement while leaving an unspecificed bequest, as an example on the other end of the spending spectrum.

We can help our clients by moving away from asking them about what they think about their probability of “fill in the blank” by giving them prudent points of references such as information from any period life table.** A practitioner’s deeper understanding of the impact of cash flow on portfolio values (how cash flows are “metabolized” … see next paragraph), as well as the inter-connectivity between the two, also helps retirees make good decisions about the prudence of their retirement income at any given point in time and age.

For the client, their concern is prudent income for the coming years. Physicians understand how different medicines are metabolized differently by people as they age. They deal with individual uniqueness as to physiological conditions and uncertainty of the progression of an illness or disease as a result of each persons state of health and age. They prescribe in terms of benefits and uncertainty as to side effects of medicines or procedures in their patients medical lives. Planners can do the same for their financial clients’ lives.

All the above dynamics are captured and reviewed each year and is why an annual checkup is important.

More technical details for the interested advanced readers:

#Exponential rates are found in the Required Minimum Distribution IRS tables and the exponential nature is an effect that becomes more pronounced for older retirees who have shorter lifespans remaining. Exponential growth happens to the withdrawal rate, while exponential decay happens to the portfolio balance. The Dec 2012 paper applies a method to mute the exponential nature of this kind of distribution so it becomes more linear and therefore extends portfolio balances into centenarian ages as needed. One never gets to the end of the distribution period because, a new annual calculation replaces the calculation of the prior year recognizing that *expected* longevity age is always older than your current age (i.e., you never catch your expected longevity age). Thus, the model more closely reflects what happens in real life with annual checkups.

*The Nov 2011 paper describes that a rising percentage of failure rate is in fact an early signal that spending should be trimmed by just a bit. The earlier a spending adjustment is made the better since the key to sustainable cash flow, our research shows, is portfolio value preservation. When spending adjustments are made early, the magnitude of the cut may be reduced as compared to waiting longer to make the change. You may think that changing allocation as a result of bad markets would do this too, however the same paper (Nov 2011) shows this is not effective (portfolio values have already gone down so it’s too late). The glue that holds it all together is the cash flow from one year to the next. Spend a little more in an early year reduces the portfolio balance which reduces the cash flow for a later year. Poor market sequences may result in inadvertently spending more than intended relative to the portfolio value. A rising percentage of simulation failures signals this problem. Alternatively, a falling failure percentage signals an ability to scrape dollars off the portfolio for “extra” spending during those good market sequences.

**Some may quibble over Social Security’s general population table, versus the Annuity table (generally a healthier population – thus older longevity ages), versus the IRS’s table, versus the CDC’s table. They all converge around age 100. Slight older expected ages leads to slightly longer distribution periods, thus slightly lower withdrawal rates given the same allocation and target probability of simulation failure rate. Therefore, the effect of using a table with older expected ages essentially conserves the portfolio values due to the lower withdrawal rates that result from slightly longer distribution periods.

##Please reference research papers working papers at SSRN for specifics on construction of this graph. Methodology differences:

1) The methodology used in these linked papers applies real dollars throughout the simulation periods (for each year: annual real return = annual nominal return – annual inflation). Applied in actual life, the inflation actually experienced each year is captured as the retiree ages and applied during the annual review process. In contrast, research methodology by others is based on nominal dollars that start with an initial dollar distribution amount, and then apply a constant inflation adjustment to that initial dollar *value*. Example: an initial $40,000 per year the first year, then increased for 3% inflation to $41,200 the second year, then to $42,436 the third year, etc. This other methodology results in cash flows not being tied at all to the underlying portfolio values beyond year one.

2) Each year’s withdrawal rate is individually calculated through *first *setting the length of the distribution period using a target longevity percentile combined with a target percentage of simulation failure rate, and the resulting withdrawal rate is applied to that year’s portfolio balance *only*. This approach helps keep cash flow directly connected to portfolio values and allows for inflation-adjusted increases in spending year by year (Mar 2012 paper).

3) A single Monte Carlo simulation today only provides information about the long term feasibility based on the conditions present today. It does not predict which course, or which of the many thousands of simulated conditions, the future may actually take. The Model Methodology in the linked papers annually re-calculates, and serially connects, simulations so that the effect of spending decisions and their timing are more transparent and the ripple effects of spending may be observed from beginning to end. Through annual updates using new simulations based on facts as they are known then, one is able to continue to re-determine feasibility in the future. It is a step by step process through the years ahead. Smaller adjustments may be made when reviews and decision rules are made annually as compared to waiting longer before making decisions based on conditions that inevitably change with time. It is thus easier to see that portfolio values decline faster at an earlier (or later) longevity age, a longevity age that is more likely to be outlived (or not outlived), with different spending patterns (cash flows); in other words the duration of the portfolio can be translated into ages, which is easier for a person to relate to, than it is to relate to probabilities of the markets alone without any reference to what age one might be should that something happen.

4) The model is three dimensional (3-D) by including allocations from 0% stocks (100% bonds) to 100% stocks (0% bonds) and illustrated in the Nov 2011 paper.

Many try to discern differences in methodology by looking at examples for *younger* retirees. The work above explores transition from younger into *older* retirement ages and the effect this has for those who may continue to live to become centenarians.

See Part II here just-what-should-an-annual-checkup-do-for-you-during-retirement-part-2

Michael Kitces has another excellent blog on getting a grasp on what the academic terms Probability of Failure (Success) mean. http://kitces.com/blog/archives/537-Renaming-The-Outcomes-Of-A-Monte-Carlo-Retirement-Projection.html

I would say the terms Possibility of Adjusting (Spending) and Possibility of Excess (outliving your money, or passing it to your heirs) are better understood by you, the reading general public.

Another great Kitces blog discussing life expectancy probability

http://www.kitces.com/blog/life-expectancy-assumptions-in-retirement-plans-singles-couples-and-survivors/

The above is an example of what I call “Dynamic Updating.” Dynamic means making changes each year based on the facts of what is happening. There are two components that are dynamically updated:

1) Longevity: and longevity itself has two components: a) the fact that you are a year older means your expected longevity has changed as explained above, and b) the longevity themselves change over time (people are living longer today as compared to years past).

2) Returns data characteristics: returns data is updated each year based on what happened in the various market dimensions (stocks: domestic & international; bonds: domestic & international; small cap vs large cap; value vs growth; etc). The new data adjusts the historic returns and standard deviation which are then incorporated into the current year’s monte carlo simulations.

Both of the dynamic updates make small adjustments to feasible and prudent withdrawals for spending purposes. Updating projections helps keep retirement income within prudent parameters.

Here is another great longevity calculator “Actuaries Longevity Illustrator.”

http://www.longevityillustrator.org/