Wednesday, February 22, 2023

Improving Retirement Planning by Employing Basic Actuarial and Financial Economic Principles

In our post of July 23, 2020 entitled, “How to Fix Advisor Retirement Planning Models,” we summarized some of the problems with commonly used Monte Carlo approaches used by financial advisors to develop plans for retirees and near retirees. These problems were identified by three retirement thought-leaders (Michael Kitces, Michael Finke and David Blanchett) in a panel discussion at the Engage 2020 virtual conference. 

Subsequent to that conference, each of these gentlemen (and a fourth thought-leader, Wade Pfau) have proposed approaches to improve current planning practices. Some of these proposals have been discussed in the following posts:

  • March 7, 2021—"Yes, ‘Probability-of-Success-Driven Guardrails’ is a Good First Step”
  • November 28, 2021—"There’s a Much Simpler and More Robust Financial Planning Tool for Retirees Than a ‘Risk-Based Guardrails Model’”
  • June 11, 2022—"Retirement Researchers Discuss Retirement Planning Approaches”, and
  • August 10, 2022—"The Retirement Researcher Constructs a Household Balance Sheet Using Basic Actuarial and Economic Principles”

While all of the proposed improvements anticipate more dynamic planning processes (i.e., they anticipate plan changes will be made in the future if necessary), the proposals from Pfau, Finke and Blanchett also significantly rely on the use of other basic actuarial principles, including calculation of present values of household assets and spending liabilities, creation of a household balance sheet and matching of present values of non-risky assets with present values of essential expense spending.

Dr. Blanchett’s proposed approach

Most recently, Dr. Blanchett released, “Redefining the Optimal Retirement Income Strategy.” The dynamic models anticipated by Dr. Blanchett in this paper also rely on basic actuarial and financial economics principles, and I believe, for the most part, are quite similar to the planning approach advocated in this website. His approach, like ours, calls for periodic valuations of household assets and liabilities, consideration of changes in spending liabilities (and/or changes in assets) whenever the household funded ratio (or Funded Status) falls outside of certain parameters and matching of non-risky assets to expected essential expenses. Blanchett describes his dynamic models as follows:

“First, we decompose the retirement spending goal (liability) into two separate goals: needs and wants, which reflects the fact that retirees typically have varied levels of elasticity (or required certainty) associated with different types of expenditures. For example, spending on travel is generally more flexible than spending on healthcare. Second, we introduce a model where spending levels (i.e., portfolio withdrawals) evolve throughout retirement based on how the retiree’s funded ratio (i.e., financial situation) changes over time. This approach can explicitly incorporate nonconstant cash flows, which is a key weakness of most existing approaches.”

Based on our understanding of Blanchett’s approach, we believe the first two steps (or models) described above are at least approximately functionally equivalent to our Actuarial Financial Planner model and actuarial valuation process.

Blanchett also presents a third model which is designed to be used in combination with the first two models. It appears to be a stochastic simulation of future spending under the two models described above based on stochastic assumptions about the future, and it produces a “goal completion score” based on utility measures and the results of the stimulation. It is not an ongoing planning tool, but rather a tool designed to facilitate user decisions based on stochastic modeling of the proposed process. We are not convinced that this third step (model) will add the same level of value as the first two steps, and as discussed in our post of January 15, 2023, we remain unconvinced that ongoing financial planning is significantly enhanced by the use of stochastic projections of future experience with or without utility functions. 

We agree with Blanchett, however, that periodically measuring the household balance sheet Funded Status using basic actuarial principles and quantifying actions to consider with respect to household assets or household spending liabilities whenever the household Funded Status becomes too small or too large is a significant planning improvement over current practice. 

We also agree with the Kitces.com summary of the importance of Dr. Blanchett’s paper in their post of February 17, 2023 where Adam Van Deusen says,

“Ultimately, the key point is that when incorporating spending flexibility into a plan, advisors can find that sustainable spending levels for clients in retirement can be higher than when assuming a fixed spending rate. So whether an advisor is incorporating risk-based guardrails or another strategy that incorporates potential spending reductions, the key is to understand each client’s tolerance for potential spending reductions and communicating clearly the implications of future changes in their portfolio value for their sustainable spending!”

Key Take-Aways

The following are key take-aways when using either the AFP or Dr. Blanchett’s approach for planning during retirement:

  • Use of a seriatim (or dynamic) process, which we call the “annual actuarial valuation process” (which itself is basic actuarial principle), that automatically adjusts for actual experience is probably more important to successful financial planning in retirement than the actual models used in each year’s valuation.
  • Ongoing planning in retirement doesn’t need to be complicated. You don’t need to run thousands of simulations. You should annually calculate your Funded Status, and you need to be willing (and aware of the need) to make adjustments to your plan when your Funded Status becomes too low. You can (but do not have to) make adjustments to your plan when your Funded Status becomes too high.
  • If calculated properly, your Funded Status will automatically reflect actual experience as it emerges, including actual:
    • Investment returns
    • Longevity
    • Inflation
    • Spending
    • Changes in Social Security or other sources of income
    • Changes in model assumptions
  • Because the process is self-correcting over time, it is not as critical for assumptions about the future to be as accurate as in a typical Monte Carlo model where a 90% or more probability of success is desired.
  • Given the variability of the above actual experience items (and other factors), it may be difficult to generate meaningful simulations of future Funded Status projections using reasonable stochastic assumptions. However, given the self-correcting nature of the process, this potential limitation should not decrease the sustainability of the household plan.
  • There are several levers that AFP users or users of Dr. Blanchett’s models can employ to reflect a lower tolerance for potential future spending reductions, including:
    • Using more conservative assumptions about the future to determine household Funded Status
    • Building up a larger Funded Status
    • Classifying more expenses as “essential” (or “needs” as used by Blanchett)

Summary

In our opinion, moving away from focusing on achieving a 90% probability of success in a typical Monte Carlo model toward annual valuations of household Funded Status using basic actuarial principles and acknowledging that plan changes may be required in the future if the household Funded Status falls below certain thresholds is a big improvement in financial planning for retirees.

If you want to read more about the AFP and focusing on Funded Status rather than probabilities of success when planning in or near retirement, you should check out our latest articles.

Focus on Client Funded Status, not Probability of Success - Articles - Advisor Perspectives

Planning for Inflation in Retirement (soa.org)