Tuesday, May 1, 2018

The Search for Certainty in the Uncertain World of Personal Retirement Financing

This post is a follow-up to our last few posts discussing the pros and cons of stochastic modeling vs. the Actuarial Approach (including deterministic calculation of your Actuarial Budget Benchmark, annual valuations and periodic scenario testing).  Following our friend Dirk Cotton’s post of March 30 and our response post on April 6, we engaged in a number of very thoughtful and informative email discussions with Dirk on this subject, and Dirk added his post of April 23 outlining the limitations of simulations.  The substantive personal financing issues on which we and Dirk agree far outnumber the issues where we may have different viewpoints.  The purpose of this post is to bring this interesting discussion to a close on our end (at least for the time-being).  In order to avoid repeating a lot of what has been discussed in the recent posts, we will:

  • Briefly summarize the stochastic model vs. Actuarial Approach discussion.
  • Encourage you (or your financial advisor) to employ both models (or combine them) to facilitate better (more informed) financial decisions. 
  • discuss new EBRI research and its implications, and
  • attempt to tie these seemingly unrelated topics together.

Stochastic Modeling vs. the Actuarial Approach with its Deterministic ABB 

As discussed by Dirk in his most recent post, “Monte Carlo simulation can be a powerful tool for retirement planning because it provides more information than other approaches.”  We agree.  MC modeling can provide probabilities of maintaining spending at desired levels, probabilities of accumulating desired amounts of assets at death, probabilities of other outcomes, volatility, etc.  All of these probabilities and projections of the future, however, depend on the reasonableness of assumptions for future real returns and variances for various classes of assets and a number of other “simplifying assumptions”, that may or may not be reasonable.  Therefore, we believe one should be skeptical about the ability of a stochastic model to realistically model the future.

The Actuarial Budget Benchmark (ABB) won’t give you these probabilities and it doesn’t model the future like MC modeling.  It is more of a dynamic spending budget algorithm than a model of the future.  It uses real (not assumed) current annuity market pricing information to answer the question, “if I were to buy annuities and insurance today for all my future spending liabilities (including desired bequests), how much would I be able to spend this year?  In other words, how much would it theoretically cost to defease (or settle) my future spending liabilities with a relatively low risk investment strategy.  Note that we are not suggesting that you actually go out and buy annuities (although you could) but rather that you go through this theoretical annuity pricing exercise to see what you could afford to spend this year if you did.

We get it.  Monte Carlo modeling is very seductive.  All you have to do is trust the underlying real return and variance investment assumptions and spending algorithm built into the model, run 10,000 simulations and the model will provide you with a wealth of information you won’t get from a deterministic model.  We aren’t going to go into all the limitations of Monte Carlo modeling in this post.  You can read Dirk’s excellent April 23rd post for his list of the limitations, with which we agree.

If you do rely solely on a Monte Carlo model, we encourage you to make sure that:

  • The model adequately considers all your assets and all your spending liabilities, including expected uninsured long-term care costs and unexpected expenses, not just recurring annual expenses.
  • You are comfortable with the expected return and variance assumptions and the spending algorithm (which typically assumes that you will spend $X each year in constant dollars for the rest of your life) built into the model.
  • The model develops a probability of having your total spending (not just your withdrawals from your investment portfolio) fall below some desired recurring spending level (like your essential expenses).
  • You revisit the model periodically to adjust for actual experience or changes in your personal situation.  Just because the Monte Carlo simulation gives you a probability of success over the remainder of your expected lifetime, don’t assume you are set for life without future adjustments.
  • You separately test various potentially significant alternative scenarios (what if testing) to help you develop contingency plans.  

Benefits of Using Both Models in Combination

Instead of choosing one approach to the exclusion of the other, we suggest that you (or your financial advisor) employ both approaches for the following reasons:

  • Since it is based on real market information for relatively low-risk investments, the ABB can be used to properly calibrate the assumptions used in the Monte Carlo model for more risky investments.  While investment in equities and other higher risk investments is expected to yield greater returns, actual returns on those risky assets may be much higher or lower than low-risk investments.  If you are invested in risky assets and you would like to have a 90% probability that your future recurring spending won’t fall below $X per year, you should expect that X will be less than your ABB (with all things, like long-term care cost reserves, etc., being equal).  If it isn’t, your Monte Carlo real rate of return or variance assumptions may not be reasonable.
  • If you invest in risky assets and the Monte Carlo model indicates that your sustainable spending level is significantly higher than your ABB (or your financial advisor indicates that you can significantly increase your sustainable spending level by increasing your investment risk), this can also be an indication that the Monte Carlo modeling assumptions may not reflect investment risk properly.
  • If the current spending level produced by the MC model differs significantly from your ABB, it is a worthwhile educational exercise to discover why this is the case.
  • Comparing your annual Monte Carlo spending plan with your ABB over time can also keep you from going too far off the actuarially balanced mark to annuity market track and can be useful in helping you develop a plan of action if you do (as discussed in our previous post).
  • Lastly, you could ask your financial advisor to incorporate the ABB spending model directly into his or her MC model as a variable spending algorithm, so that the model assumes future spending will follow the more robust ABB model (or a smoothed version of the ABB model) rather than the typical approach of assuming that you will spend a constant real dollar amount each year.  

But, enough discussion (for now) about models that attempt to tell you how to maximize your current spending and still accomplish your long-term retirement goals.  We will now switch gears to talk about how retirees actually spend, their apparent desire to preserve assets in retirement and implications for your retirement planning.

EBRI Research Confirms Many Retirees Don’t Decumulate Their Assets

Recently released research from the Employee Benefits Research Institute (EBRI) confirms prior research from a number of sources, including the Society of Actuaries, that many individuals and couples don’t appear to spend down their assets in retirement.  In fact, the research indicates that, “While some retirees do spend down most of their assets in the first eighteen years following retirement, about one-third of all sampled retirees had increased their assets over that period,” and there appears to be a strong evidence showing that many retirees preserve assets in retirement. 

There are many possible reasons why retirees don’t decumulate their assets, including:

  • They want to leave assets to heirs.
  • They worked hard to save these assets prior to retirement, and they don’t want to see them reduced.
  • They are fearful of what the future holds and feel safer and better prepared for unexpected contingencies by preserving their assets.  We call this the fear of future uncertainty.
  • They are content with their current level of spending and don’t desire to increase it.
  • They like being frugal and how preserving or growing their assets makes them feel (more in control).
  • They spend their income and view their accumulated savings as a nest egg not to be touched.

There are lots of valid reasons why you may want to use the “Maintain Your Principal” Strategy that we discussed in our post of January 14, 2018, or even grow your assets in retirement.  Will Selden, our friend at the very entertaining (especially if you are into the mathematics of retirement financing served in a large bowl of humor) blogpost RiversHedge is not ashamed to admit that he is a “Grubby Miser” and likes being frugal.  If you (or both you and your spouse) are Grubby Misers, that is fine with us.  On the other hand, if either you or your spouse are unhappy in retirement and would rather be spending more but are afraid to (i.e., you suffer from the fear of future uncertainty), you may want to look into developing a reasonable spending budget designed to increase current spending and spend down your assets in retirement, including if possible, allocating specific amounts in your budget for the things you want to do, such as travel or other “non-essential” expenses. 

But Can You Be Certain That You Will Always Have Enough to Meet Your Essential Expenses?

If you chose to spend down your assets in order to increase your current spending, you will generally be increasing your risk of having to reduce your future spending (unless you are very rich or you buy insurance, including life annuities, to cover every contingency).  For the rest of us who are not rich, who do not buy insurance and who invest in equities, we need to find the right balance of being too conservative or too aggressive with respect to our assumptions about the future and our spending.   We believe the Actuarial Approach, including calculation of your ABB, can help you find the right balance.  And yes, we also believe that reasonable Monte Carlo modeling can also be helpful in finding the right balance, and, as discussed above, combining the approaches can be even more helpful than using just one approach.  

The bottom line, however, is that the only ways to achieve certainty in an uncertain world of personal financial planning are to purchase insurance for these contingencies or to significantly underspend.   The research discussed above shows that retirees appear to be more willing to underspend than to fully insure and annuitize.   If you suffer from the fear of future uncertainty, we don’t think you should necessarily significantly underspend, but should make reasonable assumptions about the future and budget accordingly.   We also believe that even if you don’t want to maximize your spending (and you like being a Grubby Miser), it is helpful to know approximately how much you could afford to spend each year and still meet your financial goals.  We also believe it is important for you to be flexible with your spending and be willing to cut back if experience is less favorable than you assume (or increase if you have been too conservative). 

Limitations of the Actuarial Budget Benchmark

Is the ABB perfect?  No.  Your ABB may be too conservative because, for example:

  • It may understate your actual future investment returns (since you invest some or all of your assets in equities and expect higher returns than you get from life annuities).
  • It may overstate your future lifetime.
  • It may overstate your future uninsured long-term care costs (only about half of retirees spend any time in a nursing home or in assisted living).
  • It may overstate future inflation.

Or, Your ABB may be too aggressive because, for example:

  • Even though you invest some or all of your assets in equities, your actual investment returns may be less than what you can achieve with life annuities.
  • It may understate your future lifetime.
  • It may understate your long-term care expenses as some individuals spend longer than the average periods of time we recommend for ABB calculation in nursing homes or in assisted living facilities. 
  • It may understate future inflation.

And while all the above items are unknown and can cause fear of future uncertainty, we find that the elephant in the retirement planning room that very few individuals want to discuss (or even think about) is long-term care.  Insurance for long-term care is generally difficult to purchase or very expensive.  Also, there is a huge variation in actual costs for individuals, so future costs can be difficult to predict (and therefore a significant source of fear of future uncertainty).  These unknown long-term care costs make retirement planning difficult, but not impossible, in our opinion.  The ABB considers the cost of long-term care and recommends assuming average costs.  You can, of course, use more conservative or more aggressive cost assumptions for this potential expense if you want, but we believe it is important for you to address your potential long-term care costs in some rational manner. 


No one (not even retired actuaries or financial advisors) knows what future investment returns will be, what your future long-term care expenses will be or how long you will live.  This makes retirement planning difficult.  We do know, however, what insurance companies will charge for annuities (and long-term care insurance where available).  We believe this current pricing information can be useful in helping you develop your retirement plans and budgeting.  If you don’t purchase insurance to cover your future spending liabilities and you chose to self-insure, you can’t be certain your assets will be sufficient to cover those liabilities.  If you are happy being a “Grubby Miser” and maintaining or growing your assets in retirement, that is just fine with us.  However, if you believe you would be happier increasing your current spending and you are willing to live with a little more uncertainty, you might want to try out a Monte Carlo modeling approach or the Actuarial Approach (and ABB), or even better, a combination of these two approaches to help you make better financial decisions and achieve your retirement goals.