Thursday, February 23, 2017

The Actuarial Approach and the Importance of Ongoing Financial Planning

Many financial advisors utilize Monte Carlo analysis (MCA) to help their clients develop financial plans in retirement.  We have written in the past about the potential problems of using MCA, and frankly we are not big fans of relying on it exclusively.  The Actuarial Approach that we advocate utilizes transparent deterministic assumptions and anticipates ongoing (generally annual) valuations of a retiree’s assets and liabilities to help keep a retiree’s spending on track throughout retirement.  Because it does not use MCA and appears to be more volatile than the approach they use, the Actuarial Approach is looked upon by many academics and financial advisors as somehow inferior. This post will once again:
  • attempt to defend the approach we advocate as just as good, if not better than MCA, and 
  • encourage individuals and their financial advisors to consider using the Actuarial Approach for financial planning, at a minimum as another data point to be considered in the spending decision process.
The inspiration for this post was a recent Michael Kitces’ blog post written by Derek Tharp, which set forth steps financial advisors can take to avoid having their clients misinterpret their MCAs.  We agree with Mr. Tharp that there are several potential problems with MCAs and some of these problems can be mitigated if financial advisors:
  • stress that “Clients should understand planning is not a one-time occurrence”, 
  • “Emphasize the importance of ongoing planning”, and 
  • “Present information in more than one way.”

We believe financial planning is a process that benefits from periodic attention.  As pension actuaries in our former lives, we performed annual actuarial valuations to determine annual contribution ranges for our plan sponsor clients.  Like the process anticipated by the Actuarial Approach, the pension contribution determination process involved periodic measurements of assets and liabilities, along with deterministic assumptions about the future.  We and our clients both knew that the assumptions we made about the future in a pension actuarial valuation would not be exactly realized in subsequent years, and the plan’s future annual contribution ranges would change somewhat from year to year as actual experience emerged.

This pension actuarial process was not then and is not now a “set and forget” process.  We did not, as a general rule, do a MCA with 10,000 simulations of the future, tell our clients that keeping this year’s contribution level and the current asset mix fixed in future years had a 92.3% probability of successfully funding the plan for the indefinite future, and leave it at that.  It was understood that there would be ongoing valuations and changes to keep the plan’s funding on track.  The client also understood that there was considerable contribution flexibility built into the process, as the impact of future experience deviations and changes in assumptions on future contribution ranges could be smoothed to some degree.  It is with this same “deterministic assumption and annual valuation process” background that we approach personal financial planning.

Monte Carlo Analysis

Monte Carlo analysis (or Monte Carlo modeling) attempts to forecast the future based on historical experience.  This is somewhat analogous to trying to drive a car while looking out the back window.  There is an excellent likelihood that future experience won’t be anything like prior experience, and the projection will be inaccurate.  Running 10,000 simulations does not improve one’s ability to forecast the future.  Under MCAs used by many financial advisors, historical real rates of return and probability distributions of returns for various asset classes are assumed to continue.  For the client, MCA is a non-transparent process that requires a fair amount of faith.

To make these projections somewhat more realistic, some financial advisors adjust historical returns to reflect current economic conditions.  Since the primary output of a MCA is a probability of success for a given level of real dollar spending, there is an implication, if the probability of success is high enough, that the resulting spending level is essentially guaranteed and need never be changed. To achieve this result, the analysis assumes not only that historical returns will repeat themselves, but that the client will spend exactly the specified real dollar spending budget each and every year in the future.  Mr. Tharp is correct that clients may be easily mislead by MCAs.  What the clients do know is that MCAs involve lots of sophisticated calculations, so they figure they must be right.

Monte Carlo Analysis vs. the Actuarial Approach

By comparison, the Actuarial Approach (utilizing recommended assumptions) assumes future deterministic investment returns based on current insurance company annuity pricing.  These investment return assumptions are independent of the client’s actual investment strategy.  If the client’s actual future investment returns deviate from this assumed rate, the assumed rate is changed or if actual spending deviates from the spending budget, the client’s future actuarially determined spending budget will increase or decrease accordingly.  This does not imply, however, that the client’s spending must fluctuate from year to year, as the client’s annual spending budget (or actual spending) can be smoothed to some degree.

It has been our experience that an initial spending budget for a retiree who desires a relatively high probably of success under a well-conceived MCA approach that properly recognizes all sources of income and all significant expenses, is generally comparable to the spending budget developed under the Actuarial Approach with recommended assumptions.  In fact, the Actuarial Approach may produce higher initial spending budgets than the MCA approach under these circumstances.  It has also been our experience that initial spending budgets developed using adjusted historical experience and high probabilities of success don’t vary greatly based on the client’s asset mix, as the higher expected returns expected from mixes containing more equities are mostly counterbalanced by the larger amount of risk in such investment portfolios.

While initial spending budgets developed by the two approaches may be comparable under certain circumstances, the Actuarial Approach offers several features that are not generally available under a traditional MCA:

  • It allows one to easily model investment risk and spending risk.  Our workbooks contain a 5-year projection tab that gives the client the opportunity to model the effect on future actuarial spending budgets of deviations in future investment returns and spending.  This type of information can be helpful in developing investment strategy and general financial planning. 
  • It allows one to model different future spending patterns.  Unlike MCAs which typically assume constant real dollar future spending, our workbooks permit the user to assume declining real dollar future spending more consistent with observed spending in retirement.  The budget by expense-type tab in the ABC for Retirees also permits the user to make different increase assumptions for different types of future expected expenses.

Monte Carlo Analysis and the Actuarial Approach Can Work Together

If the MCA properly considers all the client’s assets and future expenses/liabilities, uses reasonable assumptions about the future, and the client is comfortable with a given probability of success and constant real dollar spending in retirement, the MCA approach may produce a reasonable spending budget for the client.  As Mr. Tharp says in his article, however, it is important for clients to recognize that developing a spending budget is not a one-time event and should be revisited periodically.  In addition to reflecting actual investment performance and actual spending, the client’s spending budget may also change over time as the client’s spending goals change.  We believe the data points obtained by applying the Actuarial Approach on an annual basis can:

  1. Be an independent check on the reasonableness of a Monte Carlo analysis, 
  2. Be an important supplement to the data points developed by a MCA in keeping client spending on track and consistent with the client’s financial objectives, and 
  3. Present information in a different way to increase client understanding.
Therefore, we encourage retirees and their financial advisors to periodically compare the spending budgets they develop with their MCAs (or other approaches) with spending budgets under the Actuarial Approach.

We like the way Mr. Tharp thinks, and we look forward to future articles by him.  In a future blog post, we will discuss how Mr. Tharp’s post of February 22 regarding development of spending budgets that are expected to decline in real-dollars as retirees age is yet another advertisement for using the Actuarial Approach. 

Saturday, February 4, 2017

Five Ways to Increase Your Near-Term Spending, Part II

This post is a follow-up to our post of November 30, 2015 in which we talked about ways to increase your near-term spending in retirement.   In that post we discussed:
  1. Finding part-time work or other sources of income 
  2. Deferring commencement of Social Security or purchasing annuities 
  3. Using more aggressive assumptions in your calculations 
  4. Using more aggressive assumptions for non-essential expenses, and 
  5. Simply increasing your budget (or your spending) by x%
We also cautioned our readers that, all things being equal, increasing near-term spending increases the risk of declining real (today’s) dollar spending later in retirement.  In this post, we will focus on a subset of the third approach discussed above; lowering the assumed annual target rate of increase for future spending budgets to increase current spending budgets (or reduce the assets needed to fund a given level of spending). 

Within the past few years, several researchers and retirement experts have observed that retiree spending appears to decline in real dollar terms as individuals age.  We discussed this research and how retirees could use our spreadsheets to anticipate declining real dollar spending in developing their spending budgets in our posts of March 31, 2016, August 20, 2016 and November 4, 2016.  More recently, a retirement expert from the UK, Abraham Okusanya, argued in this article that spending in retirement does not follow a “U-shaped pattern” as previously thought, but rather declines in real dollar terms throughout the entire retirement period.

Considering the growing volume of research showing declining real dollar spending in retirement, several retirement experts have suggested that individuals should consider developing their spending budgets so that they also decline in real terms throughout retirement. For example, Mr. Okusanya implies that, based on spending research in the U.S., it would be ideal to target inflation minus 1% (or more) for purposes of developing future spending budgets in the U.S. 

The retirement experts have concluded that this lower target for future spending means that either near-term spending can be increased or the amount a person needs to save for retirement can be reduced, compared with assuming a constant real dollar future spending target.  The experts are less clear, however, as to exactly how much spending may be increased (or savings decreased) by assuming the lower future spending target.

As with all spending matters, we at How Much Can I Afford to Spend leave decisions of how much you spend in a year up to you and your financial advisor.  We simply provide you with tools that give you data points designed to help you make your spending decisions.  However, unlike the retirement experts, we can easily quantify for you how much your current spending budget will be increased (or your necessary savings decreased) if you assume that your future spending budgets will increase by inflation minus 1% in retirement, rather than by inflation. We determine the relevant percentages by first taking the basic actuarial equation that is the foundation for this website:

and manipulating it to obtain: 

To quantify how much This year’s spending budget will increase by targeting future spending budget increases of inflation minus 1%, rather than inflationary increases, we need to divide PV future years increasing by the assumed rate of inflation by PV future years increasing by inflation minus 1%.  The PV future year values are available in the Present Value Calcs tab of our workbooks. 

Similarly, the reciprocal of this ratio will give us the % decrease of needed savings to produce a desired level of spending.

The table above shows the results under our current recommended assumptions at various ages. So, developing a spending budget at age 65 under these assumptions and further assuming future spending budgets increase by 1% per year, rather than the recommended inflation assumption of 2% per year, would increase the actuarially calculated spending budget by 13.3% (or decrease the adjusted assets needed to provide the desired level of spending assuming retirement at age 65 by 11.8%), all things being equal.

Note that if you are, or your financial advisor is, determining your spending budget by adding the results from a Systematic Withdrawal Plan (SWP) to your income from other sources (including Social Security), it may be somewhat more difficult than as described above to develop a spending budget designed to increase at a rate other than inflation.  As discussed in prior posts, SWPs are not really designed to work well unless Social Security is the only other source of income in retirement and the retiree’s spending objective is to have constant real dollar spending in retirement.

While research may support decreasing real dollar spending in retirement, we encourage our readers to develop their future spending increase assumption (or assumptions) by separately examining expected future increases for the three types of future expenses in our Budget by Expense Type tab in our Actuarial Budget Calculator (ABC) workbooks:

  • Essential non-health expenses 
  • Essential health expenses 
  • Non-essential expenses
Since it is not unreasonable to assume that future essential non-health expenses will increase with inflation and essential health expenses may increase at a faster rate than inflation, you may not be comfortable assuming total future recurring spending budgets will increase at a rate of inflation minus 1% (or more) unless your non-essential expenses are assumed to be a relatively large component of your initial spending budget.

Thursday, February 2, 2017

Avoiding Financial Regret about Retiring Too Early

In their recent Wall Street Journal article, “Before Retiring, Take This Simple Test,” Dr. Shlomo Benartzi and Dr. Martin Weber encourage individuals considering retirement to take a “two-question quiz that can help predict whether [they will] regret the timing of [their] retirement.”  Dr. Benartzi is a behavioral economist professor at UCLA, and Dr. Weber is a professor at the University of Mannheim in Germany, with special interests in behavioral finance and its psychological foundation.  It is an interesting article that advocates the use of behavioral economics tools like the two-question quiz to “help us find ways to stop people from retiring too early" and regretting their decision.

The two questions in the quiz are almost the same, but the slight difference in timing of the two questions allows for measuring the consistency of time preference.  That is, answering the two questions inconsistently exhibits preference for immediate rewards, as opposed to postponement of rewards.

The recommendation comes from a study of over 3,000 Germans who answered the quiz.  Per the authors, the respondents with inconsistent answers to the quiz:

  • “exhibit a tendency known as present bias, or hyperbolic discounting”, 
  • “tend to retire…earlier (about 2.2 years on average) than those with consistent preferences”, and 
  • “over time, these people are also far more likely to say they regretted the timing of their retirement.”
The authors also concluded that retiring about 2.2 years on average earlier than those with consistent preferences resulted in “roughly a 13% reduction in their monthly benefits.”  Perhaps the German retirement system is different from that in the U.S., but as we will show in the example below, each year of continued employment and deferral of retirement from age 62 to age 70 results in closer to a 10% increase in an individual’s real dollar spending budget, rather than the 6% figure (13% divided by 2.2) cited by the authors.

As retired actuaries and not behavioral economists, we at How Much Can I Afford to Spend in Retirement believe that substituting facts for appearances and demonstrations for impressions are still good ways to influence individual behavior.  And while we share the goals of behavior economists to help people make better decisions, we believe that any “framing” of the retirement age decision should be based on reasonable calculations.


Let’s assume we have a single female, Beth, currently age 62, making $100,000 per year.  She has $500,000 in accumulated savings, in addition to her home equity.  She is currently eligible to receive an immediate annual Social Security benefit of $20,124, but she has no other sources of retirement income.  Her employer sponsors a 401(k) plan that matches contributions up to 6% of pay with a 50% match.  Beth contributes enough each year to receive the maximum employer match and her annual savings (including her 401(k) contributions) are 15% of pay. 

Beth wants to know about how much her real annual spending income would be if she retired today at age 62 or if she kept working and retired at age 65, 68 or age 70.  For immediate retirement, she uses our Actuarial Budget Calculator (ABC) for retirees workbook.  For the other ages at retirement, she uses our ABC for pre-retirees.  She makes the following assumptions and other data entries:

  • Our recommended assumptions for discount rate (4%), inflation (2%) and lifetime planning period (death at age 95) 
  • Her pay will increase in the future with inflation (2%) 
  • No amounts desired to be left to heirs 
  • Present value of unexpected expenses: $50,000 
  • Desired increases in future spending budgets equal to inflation (2%) 
  • Continued pre-retirement savings rate: 15% 
  • The present value of her long-term care costs will be covered by her home equity
She uses the Social Security Quick Calculator to estimate what her Social Security benefit in today’s dollars will be (shown in the table below) if she continues her employment.  She enters the following amounts into the ABC pre-retirement workbook for future dollar amounts (today’s dollars increased by 2% per year inflation):

The Table below shows the results of Beth’s calculations:

The table shows that, given Beth’s information and assumptions, her spending income will increase about 10% for each year she continues to work and save 15% of her pay.  She can either choose to look at how much her income will increase each year by working or, as suggested by the behavioral economists, she can look at her relative loss by not working.  However she looks at it, she will benefit from using our workbooks to help her make her decision about when to retire. It is important to note that results will vary for different individuals and you should always model your own situation.