# Summary: Two schools of thought on retirement income.

Retirement income management boils down to basically two schools of thought … Probability Based School and the Safety First School. Wade Pfau has a summary on his blog, and I’ve written on the comparisons as well. Wade is an advocate of the Safety First School in general while I’m an advocate of the Probability Based School.

I discuss in detail the analogy of ocean waves to visualize retirement planning in this blog post.

Rather than writing a description of the two, here’s a graphic that illustrates that both schools are talking about the same ocean (ocean analogy explained more). They simply are viewing it from different parts of the ocean.

(Please click on graphic to enlarge – or hold shift, then click graphic to enlarge in a new window).

• Notice that both graphs have retirement starting on the left side of the figures and ending on the right side of the figures.
• That numbering difference explains a number of differences between the schools of thought.
• The upper graph numbers (along the very top) from higher to lower values from start to finish of retirement – represents counting down the number of years remaining in retirement.
• The lower graph numbers (along the very bottom) traditionally from lower to higher values from start to finish – represents count the number of years since retirement began.
• Stocks in both graphs are labeled in the blue zones while bonds/cash are labeled in the black or brown zones in either graph.
• The upper blue stock zone decreases in emphasis while the bonds/cash zone increases in emphasis as one moves from left to right – represents a declining stock allocation exposure as one ages.
• The lower blue stock zone increases in emphasis while the bonds/cash zone decreases in emphasis as one moves, again, from left to right – represents an increasing allocation to stocks as one ages.
• Sequence risk may be visualized in the graphs as being represented by the wave height between peaks and troughs (this blog post discusses how to measure and use this visualization and information).
• Probability based (upper graph) views sequence risk as being more problematic at older ages (the lower numbered years – upper right). Most also don’t believe sequence risk goes away at any age – it’s ever present. Thus exposure to the risk is managed through a higher allocation to non-stock holdings; in other words, a declining equity glide path with age.
• Safety first (lower graph) views sequence risk as being more problematic at younger ages (again, the lower number years – but now, lower left). Proponents view sequence risk as going away after 7 to 10 years. Thus exposure to stocks at older retirement ages; in other words, a rising equity glide path with age.
• Most safety first proponents envision spending patterns as following the waves (portfolio value) and spend a percentage of the portfolio value as balances change up and down between peaks and troughs. This would lead to constantly changing spending amounts since the portfolio value is constantly changing.
• A more stable spending pattern may be obtained by calculating the trough portfolio value and base spending percentages on that more stable value. Balances above that trough value may be spent on a discretionary basis when conditions permit. This is a method for probability based proponents to manage sequence risk and variable portfolio values.

The difference primarily is when do they envision the “surf zone” or the low numbered years of retirement? How to address those surf zone years is not in dispute – use of Single Premium Immediate Annuities (SPIAs) and/or Deferred Immediate Annuities (DIAs). Most in the profession view application of these during the early years of retirement (bottom segment of the graph, and numbering at bottom, representing safety first view). Others, including myself based on lifetime expected cash flow analysis, view application of these as an option to be made later in retirement (upper segment of the graph, with numbering at the top, representing probability based view).

Another difference between the schools is whether stock exposure should rise or decline during the retirement years. Again, research supports either view – differences primarily due to simulation design, assumptions and data interpretation.

Differences also come in as to when to adjust calculations for the present year’s income and to what point is that income calculation in reference to … when retirement began (safety school), or from the present year through simply redoing the forward looking calculations each year?

How are actuarial lifespans calculated and applied … at the beginning of retirement using a conservative age less likely to be outlived (i.e., a safe fixed period), or each year with a new reference to the life tables to recalculate the probable time now left in retirement?

What is the reference point for decision rules about when income may be increased or possibly decreased, based on market and spending patterns … with reference to a past point (Guyton), or with reference to the present calculations’ percentage of simulations that fail the calculation (hint: higher failure rates signal spending that’s probably too high)?

Basically, the Safety First School continually refers back to the beach as its reference point (bottom view) as one makes future decisions. The Probability Based, dynamically updated, School refers to how long until one reaches the beach (upper view).

Notice that the application of insurance products that insure income such as Single Premium Immediate Annuities (SPIAs) and/or Deferred Immediate Annuities (DIAs) are applied at opposite ends of the retirement spectrum, when the numbered years are in their low values (recall that the top and bottom numbering of the graphs are in opposite directions). Those are the years viewed most at risk but they are at opposite ends of retirement because of how either views where one is at the beginning and where they end … in deep water going towards the shore? Or from the shore going into deep water? The differences emerge primarily how either school views sequence risk and how to manage it.

Thus, visually from the graphic above …

1) Counting retirement years up: is the retiree mentally moving from the beach (safety first) into deep water, and can they emotionally adjust from shore to deep water differences once they’ve aged?

2) Counting retirement years down: is the retiree mentally moving from deep water (probability based) towards the shore as they age?

Neither approach has been proven to be superior to the other, and I think neither should be because either approach may be appropriate to different client situations and having either arrow in the quiver is likely to be essential to address diverse situations and circumstances.

However, understanding the fundamental differences is important.

Note: It is unlikely one would reach very low time remaining numbers from the life tables because one always has an older expected longevity age and time frame based on current age. Explained a bit more in this paragraph from linked blog post: “Notice in Figure 4 (of linked blog) that the initial expected retirement time period at age 60 is 33 years (couple of the same age using Annuity 2000) – yet when the retirees reaches age 93, there’s still a possibility of 11 more years! The longevity table at the bottom of Figure 4 shows the percent of cohorts expected to outlive (i.e., percentile) time periods at those ages. For example, 50% of cohorts may expect to outlive 33 years at age 60, while once at age 93 20% of those cohorts may expect to outlive another 11 years (as couples). Again, life tables for individuals are different and would apply when one of the couple dies, or would be used immediately if the person were single. The time periods would be shorter, so withdrawal rate and cash flow would be higher relative to couples. Males would compare “better” to females for the same reason (do you want more money, or more life?).”

PS. I’d like to thank Dirk Cotton for his comments as I wrote this post.

### 7 Responses to Summary: Two schools of thought on retirement income.

1. Larry Frank, Sr. July 30, 2015 at 8:22 am #

Kitces has done an excellent job explaining how the 4% safe approach has done as a conservative spending methodology here https://www.kitces.com/blog/how-has-the-4-rule-held-up-since-the-tech-bubble-and-the-2008-financial-crisis/

I link to it since he also illustrates how counting the “years since retired” view works versus the “years left IN retirement” view I discuss above.

2. Terry Robinson September 13, 2015 at 8:48 am #

Very interesting piece. I was a bit puzzled by the ocean illustration. It appears that under the probabilistic approach one would hold nearly all retirement assets in stocks for the first several years and then introduce bonds and annuities. I read this and the early post describing the ocean analogy and remain confused as to what exactly the blue represents. Is it income from equities including the liquidation of shares or is it the value of the equity portfolio through time?

For my retirement plan which I hope to start in a year at age 62, I plan to delay SS until age 70 for the benefit of my younger spouse who hasn’t much of a work history. Using made up numbers as an example, I’ll need to spend say \$100,000 each year increasing with inflation and have a small pension of say \$20,000 which starts at 62 with no inflation adjustments. In the example I’m short \$80,000 annually which must be made up by drawing down other investments untill SS kicks in at age 70 (at that point let’s say one-half of the \$80k will be covered by SS). My plan is to cover \$40 to \$50 thousand of my early year shortfall via SPIAs and the balance from my stock and bond portfolio, this seems at odds with your probabilistic chart. It feels as though I have blended the two approaches a bit.

I suppose it would help me to understand how a substantial defined benefit pension plan taken as a single life annuity appears in the two charts.

Very much appreciate the work you do.

Terry Robinson MBA,CPA, CFP registrant

3. Larry Frank, Sr. September 13, 2015 at 10:13 am #

Yes, under the probability based view, one would hold more stock when younger and then reduce the stock allocation as one ages, up to and continuing through retirement. Since both views are talking about the portfolio source for retirement income, stocks would be replaced by bonds and/or cash slowly over time in the probability based view (the opposite in the safety first view). When you say annuity, if it is not annuitized, then that money is really part of the stock and/or bond portfolio depending on if the annuity is a variable annuity or fixed annuity. I won’t go into those here since there are other posts that do that.

An immediate annuity, that basically is like a pension income that you have bought, and Social Security, are not part of the ocean illustration because those income sources are not variable; they don’t go up and down with markets like stocks and bonds do. Thus, the ocean concept is just talking about the supplemental retirement income that comes from portfolio sources. This gets added to Social Security, pension, and immediate annuity income (called Single Premium Immediate Annuity). Pension and SPIA most often have issues with payments keeping up with inflation, called loss of purchasing power. The purpose of the portfolio is to provide additional income that does keep up with or moves ahead of expense inflation.

The “blue” labeled Stock is the VALUE of the stock portion of the portfolio. The bond & cash labeled area represents the VALUE of the bonds and cash. The income from both areas would be from ALL sources of money including sale of stocks and bonds. So yes, selling shares is one source of money. Other sources are from capital gains and dividends; both can come from both stocks and bonds. This is called a TOTAL RETURN approach. Money does not know its’ source, it is fungible, and all money can be used for any retirement expense.

Your example is a common mixture of ALL sources of income beyond the portfolio sources and a good example of what is called a “bridging strategy” using more portfolio sources for income during the early years of retirement and then replacing that source with a now higher Social Security (which by the way, maximizes the survivor income when waiting to age 70 to receive it. You could also file and suspend at or after your Full Retirement Age so that your spouse can claim her spousal benefit when she’s eligible, in your case assuming she’s not more than 8 years younger). The portfolio is what is represented in the figures illustrating the stocks, bonds and cash. The TOTAL portfolio value goes up and down when you look at statements and your statement total would be represented by the surface of the ocean which goes up and down.

Thank you again for your comments that does a great job clarifying that the ocean analogy is only talking about the PORTFOLIO source for retirement income. So not at odds at all. I simply didn’t include non-portfolio sources in the analogy. Social Security, SPIA and/or pension income is NOT illustrated in the figures. Those sources are not labeled, and to complete the analogy for your and others benefit, would be “ocean floor bedrock” that doesn’t shift and change over time, over which the labeled “shifting sands” of bonds and cash would lie. The blue area of figures is the more active “water” of stock. That bedrock is below the figure illustrated bonds & cash. I hope the answers helped clarify the analogy for you.

4. Larry Frank, Sr. December 28, 2016 at 8:00 am #

Never late to comment since posts are always read over different times when they catch people’s attention on something they’re thinking about now – not when I wrote it.

Your comment brings up sequence of returns, increasing glidepaths, level of wealth, and how these affect spending. They are in many ways related to each other.

You are expressing the common perception that comes from the Safety First school of thought. Here, risk is perceived to be greatest at retirement (regardless of what age a person should retire at). The years are counted up … 1, 2, 3, … 10, 15, 20, … etc. as if how long you’ve been retired matters. In reality, calculations should be the same for everyone at any given age (because period life tables provide expected lengths of time for calculations) and the years should be counted down, not up, because what is relevant to the calculations is the time period.

Sequence of returns risk:

I addressed this in the post. I also address this in this post “Why Volatility isn’t so Bad” https://blog.betterfinancialeducation.com/sustainable-retirement/why-volatility-isnt-so-bad-adviceiq/ (and the more indepth post linked to in the post). I agree that market volatility causes problems during withdrawals. The difference is that the “Probability Based” view sees the problem at older ages when there is truly less time remaining to recover from market misbehavior. Solutions to market volatility issues are the same between the two schools – the difference is the view as to when the issue becomes a problem. Is your portfolio really at complete risk of loss early in retirement? No … https://blog.betterfinancialeducation.com/sustainable-retirement/is-all-of-your-portfolio-at-risk-of-loss/

Sequence of returns risk NEVER goes away. The “Safety First” perspective gives the illusion it goes away because the portfolio balance grows from underspending. Level of wealth isn’t really a factor as you may think. It is a factor RELATIVE to what the portfolio value could support for spending over the time frame used. Safety First uses very conservative time frames (long ones unrelated to retiree’s actual age), portfolio allocations (conservative), and spending amounts (4%). What tends to happen is that money is not spent now in order to have more money to spend later. However, the probability (using period life tables) suggest that few tend to live to, or beyond, the conservative age used (either 95 or 100) … of course those in their late 80’s have a higher likelihood of outliving 95 relative to someone in their late 60’s. You should also expect to get older as you get older, referring to the incorrect static use of some age like 95 https://blog.betterfinancialeducation.com/sustainable-retirement/you-should-expect-to-get-older-as-you-get-older/ … period life tables provide more insight into how long a period calculations should be performed.

Money can be judicially measured and monitored to spend over the years you’re more likely to be alive as found in this research “Just Where Does the Fear of Outliving Our Money Come From?” https://blog.betterfinancialeducation.com/sustainable-retirement/just-where-does-the-risk-of-outliving-our-money-come-from/ and explained more here https://blog.betterfinancialeducation.com/sustainable-retirement/part-ii/

There are ways to manage sequence risk at all ages … an example is explained here https://blog.betterfinancialeducation.com/sustainable-retirement/a-simple-retirement-distribution-process/

Increasing Glidepaths:

This is also related to conservative spending early in retirement (for fear of running out of money – which I explain in the Two Part “Just Where Does the Fear of Outliving Our Money Come From?” linked to above. You see – if spending is constrained and the portfolio balance is allowed to grow, eventually the balance is larger than you need for spending needs (bequests result – typically unintentionally – unless spending is adjusted because the balance is bigger). This means there are “excess” funds and indeed in that case more risk can be taken. When you look at Monte Carlo simulation results (over the typical single fixed period the Outliving Money posts discussed) the capital balances grow quite large (at least relative to need).

The Outliving Money post research showed that even if balances are allowed to grow, or decline, and spending can be adjusted just a little, there is always money available into very old ages unless a catastrophic spending need occurs to deplete the balance. The research used “extreme spending” and still showed balances, although low ones, into old age. The working paper for the published paper, shows that spending less early “pushes” the balances for spending more later – much like a teeter totter.

Level of wealth:

So yes, level of wealth is important – but that is relative to what amount they need to sustain a prudent level of spending … which also changes throughout retirement. Because one has fewer years remaining as you age, you can spend a higher percentage for those fewer years … and it slowly increases as you age because time slow goes down … in the post here, one is nearing the shoreline and entering the choppier waters of the surf zone (less time for recovery from sequence risk – which is always present, but has more of an effect when there is less time to recover – this argues for a DECLINING GLIDEPATH … i.e., getting more conservative with age).

So, if someone oversaves, they’d have more wealth relative to their spending needs and also choices, spend more once retired, spend what they need and let the balance grow, or retire earlier which results in the excess savings being needed to some degree to finance the longer retirement period. If someone undersaves, their risk if they try to spend what they need, is running out of money before running out of time. If they can, they should work longer so the time their money is needed in retirement is shorter.

Thus, level of wealth is relative to spending needs. Balances are also relative to spending needs. Portfolio risk is relative to spending needs and time remaining for those needs. Sequence of returns risk is always present at any age – just ask someone older than you – and is also relative to portfolio balance relative to spending needs that portfolio needs to support.

Basically, based on many research projects ( http://www.betterfinancialeducation.com/larrys-contributions-retirement-research-body-knowledge ), I believe the Safety First approach is overly conservative and leads to money left on the table that could be spent during years one is most likely around to spend it. Of course, that needs to be recalculated every year so that money for the upcoming year can be determined while also reserving money as “Personal Mortality Credits” for those expected years you may still have. That end point continually and slowly moves as you age. The approach you describe is more akin to a carryover from the analog days of financial planning. The stochastic (probability of randomness) approach currently in use adds Monte Carlo, but still over a single period (analog). Casting Monte Carlo simulations each year is where planning needs to go with programming. That is not being done yet by software … however, in practice those like myself who redo the calculations each year and update all the data points, to include longevity from current period life tables, are Multi-Casting.

Yes, I’m describing an approach not often written about by those who are adhere to the Safety First school of thought, single casting of simulations, etc. A deeper understanding of shortfalls of that approach leads to the Probability Based (that is also age based rather than generic periods), approach to the problem primarily a built-in method that transitions calculations from period to another (the Safety First approach doesn’t have a coherent method to transition between periods, or is based on rules of thumbs, rather than specific calculations using specific data from specific circumstances).

In summary, I’m writing about the Probability Based School here specifically because it isn’t written about very often and is less understood as a result. I use the Ocean Wave analogy to explain the differences visually as well as to show that portfolio approaches can be managed with sequence risk in mind. The insurance industry would want you to think otherwise for early year retirement. I think those tools (immediate annuities for example) are useful, however later in retirement.

Thanks for commenting!

• Larry Frank, Sr. December 28, 2016 at 10:31 am #

Indeed, you bring up a good point that hasn’t been researched in depth. What research there has been on the question of annuities hasn’t really looked at what I call “Standard of Individual Living” (SOIL) that looks specifically at income level and wealth needs to support that.

My co-researches and I did look at the breakeven choice in general between immediate annuities and portfolios in this published paper “Breakeven between Immediate Annuities and Managed Portfolios” https://www.betterfinancialeducation.com/7th-paper-breakeven-between-immediate-annuities-and-managed-portfolios (link to paper and it’s working paper on this webpage). We didn’t look at SOIL specifically, but looked at ages instead as to WHEN it seemed to make sense.

I’m looking into the question relative to SOIL where Social Security is also factored into the matter along with income levels as we speak with the co-researcher from my 8th paper. We’re using the “Multi-casting” approach developed in that paper to parse out both age AND living standards which we’re pulling from Census records.

I know Blanchett from Morningstar is also working on the income level question too and his paper, that he asked me to comment on in draft form, will be published sometime in 2017.

So better work is going towards looking into the question and issue you raise. This said, it is always better to perform specific calculations for specific situations instead of the prevalent “rules based approach” that seems to have taken hold in the financial planning profession. We are planners and so making calculations should be second nature. Unfortunately, making those calculations seems to have been forgotten once certified!

It is not so much having a “significant” margin of safety. Rather, it is designing a process that includes no speculating within the portfolio (gets to using evidence based investment approaches applied through indexing). The process should recognize and manage the difference between Deep Risk and Shallow Risk (Bernstein Indeed, you bring up a good point that hasn’t been researched in depth. What research there has been on the question of annuities hasn’t really looked at what I call “Standard of Individual Living” (SOIL) that looks specifically at income level and wealth needs to support that.

My co-researches and I did look at the breakeven choice in general between immediate annuities and portfolios in this published paper “Breakeven between Immediate Annuities and Managed Portfolios” https://www.betterfinancialeducation.com/7th-paper-breakeven-between-immediate-annuities-and-managed-portfolios (link to paper and it’s working paper on this webpage). We didn’t look at SOIL specifically, but looked at ages instead as to WHEN it seemed to make sense.

I’m looking into the question relative to SOIL where Social Security is also factored into the matter along with income levels as we speak with the co-researcher from my 8th paper. We’re using the “Multi-casting” approach developed in that paper to parse out both age AND living standards which we’re pulling from Census records.

I know Blanchett from Morningstar is also working on the income level question too and his paper, that he asked me to comment on in draft form, will be published sometime in 2017.

So better work is going towards looking into the question and issue you raise. This said, it is always better to perform specific calculations for specific situations instead of the prevalent “rules based approach” that seems to have taken hold in the financial planning profession. We are planners and so making calculations should be second nature. Unfortunately, making those calculations seems to have been forgotten once certified!

It is not so much having a “significant” margin of safety. Rather, it is designing a process that includes no speculating within the portfolio (gets to using evidence based investment approaches applied through indexing). Process means recognizing, and managing, the difference between “Deep Risk” and “Shallow Risk” (Bernstein https://blog.betterfinancialeducation.com/sustainable-retirement/william-bernstein-on-deep-risk-shallow-risk-and-investing-for-the-long-term/) such as those I describe in my prior reply to you (Ocean Waves, Simple Retirement Method, risk of loss in a portfolio, etc).

I’m glad you’re looking for more refined insights than the simplistic 4% rule of thumb. Seek on … and more is coming!

5. Larry Frank, Sr. September 19, 2020 at 9:22 pm #

Sequence of risk concerns are similar to the concern about transitioning the “surf zone” where ocean waves meet the beach.

Safety first views this as standing on the beach and trying to make it to open sea where sequence of risk is an issue during early retirement (transitioning the “surf zone” first to get to the open sea where waves aren’t crashing and breaking anymore). This view is prevalent and can be spotted by anchoring to the event or date of retirement, and counting the years UP as I explained above: 1, 2 … 5 … 10 … etc. years IN or SINCE retirement. In other words, the point of view is anchored on the date or event of retirement.

The Probability School of Thought, as I’ve presented and researched, views the problem of being out at sea and wanting to navigate the surf zone at the end of the voyage in order to land on the beach at the end. This point of view anchors measurement anchored on longevity probability where that life table age continues to roll out ahead of you, AS YOU AGE (it is NOT a fixed age) like the bow wave of a boat (you only catch it when you stop). This view counts the years DOWN (not UP) as I explained in a comment in a post linked to below: “Is All Your
Portfolio at Risk of Loss?” https://blog.betterfinancialeducation.com/sustainable-retirement/is-all-of-your-portfolio-at-risk-of-loss/