Tuesday, November 19, 2013

Variable Spending Strategies

My new Advisor Perspectives column on variable spending strategies in retirement is now available. It's called, "New Research on How Much Clients Can Spend in Retirement." In the column, I review the variable spending strategies developed by:

Guyton and Klinger
Frank, Mitchell and Blanchett
Blanchett, Kowara, and Chen

Working on this column has inspired me to try to develop my own contribution to this area. It's a work in progress still.


  1. Your article is an excellent summary of the probability based school of thought approaches.

    I look forward to your work on integrating the safety first school of thought with the probability school through your intriguing concept of an efficient frontier, that hopefully builds in the actuarial adjustment for time remaining at any given age.

    What I'm seeing from all the various research perspectives is that age tends to change the "solution" over time. One solution mix at a younger age tends not to be the same solution mix at a later age. Your work on melding both schools of thought together will be interesting to follow.

  2. "Working on this column has inspired me to try to develop my own contribution to this area. "

    With your creative thinking and your readers' (and Bogleheads) helpful suggestions and critiques, odds are pretty high that you will break substantial new ground in the area. All I can say is that I'm grateful you share your work, warts and all. Thank youl

  3. In your AP column on page 5 you mention “Each approach indicates how much can be spent, which is nice to know, but this budget is not otherwise connected to any lifestyle‐spending goals or minimum needs.”

    An achievable goal of all advisors is to have a quarterly/annual meeting with their client and provide the following statements:

    (Planning Horizon Statement) Looking at current actuary tables, a person of your age today can be expected to live to age Z years old. This implies your planning horizon is X years. For your spouse, the tables indicate…

    (Peer Outlook Statement) Looking at historical data provided by a network of financial advisors, a person of your age usually spends $$,$$$ annually to sustain a comfortable lifestyle. We note that recent retirees travel more and more senior retirees have greater needs to cover healthcare costs. On average, a person/couple of your age spends A% of their annual spending on fixed living costs (home, food, etc.), B% on discretionary costs (gifts, travel, etc.), C% on healthcare costs, etc.

    (Required Minimum Distribution Statement) The Internal Revenue Service requires that you withdraw $,$$$ this calendar year.

    (Recommended Allowable Withdrawal Statement) After reviewing your income assets (Defined Pension Plan cost of living (COLA) increases, CD/Bond Ladder rate changes, Equities performance, Whole Life Insurance withdrawals, Annuity disbursements, etc.) over the past twelve months and considering a 30% failure rate of reaching your goals, you can safely withdraw N% ($$,$$$) this year from your retirement portfolio. Your overall income this year will be $$,$$$, which is in line with your peers.

    (Tax Implication Statement) If you withdraw from your Roth/Traditional IRA, your tax implication will be… For every dollar over $$,$$$ (lower income tax threshold), you will be taxed at X%. If you make donations to charity to reduce the income to $$,$$$, then you will remain below this threshold.

    (Recommended Asset Allocation Statement) To maintain a greater probability of reaching your retirement goals, we recommend an asset allocation for your portfolio of…

    Having a safe withdrawal rate is nice to know, but how does it affect the client should always be the question. It is a troubling concept to inform the client how they are doing compared to the Joneses (their peers), but it is more comforting to know that they are making the right choices.


    1. Paul,

      Thank you, it's a great outline.

      I'm wondering about the "Peer Outlook Statement." Is this something you are already able to do? I'm wondering, because it is exactly the thing that inStream solutions will start being able to provide over the next couple of years based on individual data and not something like averages from the Consumer Expenditure Survey.

    2. Is it something I am able to do? No, I am not a financial advisor. I hope I did not mislead you. I am an engineer and a mathematician. I work for the government. My only financial tools are what I learn from the internet, such as your helpful blog (thank you).

      As for the concept of “Peer Outlook”, it has been a wish of mine for quite a while. I wish advisors would compile and submit general client data to a central collection point (CFP.org ???). The data would include an annual break-down of their clients’ age, planning horizon, total retirement asset value, and spending by category. The collection point could then turn around and provide back to all advisors a trend in spending broken down by age group (or horizon planning ???). The advisors would then use that set of trends to meet with their clients and explain how they compare to others in their similar situation (i.e. peer; same age or similar expected life remaining).

      It would not benefit me if my advisor presented me a comparison against some of his/her more wealthy clients as we are not in the same category. They are looking more toward estate transfer of assets whereas I am looking more toward retiring comfortably without running out of money. Clients in differing categories have different concerns.

    3. Paul,

      I assumed you are a planner. Your outline is very sophisticated.

      About the peer outlook, this is something that inStream is moving toward now. I'm working with them partly to help analyze the data they get. This is new financial planning software for advisors. The Peer Outlook is what they mean by "contributory knowledge" and "crowdsourcing best practices" on their webpage http://instreamwealth.com/

      Yes, I think they can do everything you mentioned, except I'm not sure yet if there data on spending by category will be good. I think they will just have something like what you see at mint.com, which I find to not be very helpful. Maybe most people will not go through the effort of correcting all of the spending classifications, so the spending data can't be trusted. But we will be able to get trends on overall expenses along all of the avenues you mention. It's moving in this direction.

      Again, congratulations on having such a well thought out list. It made me assume you are an advisor with years of experience.

  4. Hey Wade, when you judge dynamic spending approaches, don't forget to include mine, the actuarial approach actually developed by an actuary.


    And my proposed approach does not "leave out a role for income annuities."

    When testing it, feel free to use the recommended assumptions and algorithm for adjusting for actual experience discussed in my October 11 post.


    1. Ken,

      Thanks for the reminder, I'll be sure to have another look.

  5. Although a lot of energy and research has been devoted to various withdrawal strategies through Monte Carlo analysis and other methods, I'm still waiting for any of these to be integrated with realistic probabilistic models for major annual *expense* variations that retirees are likely to incur. It's much too simplistic to assume that annual expenses will increase at some constant inflation rate or that major -- and hard to plan for -- expenses won't hit somewhere along the way. How about incorporating expenses -- along with probabilities at specific ages -- for major life and health events, beyond what is typically insured against through LTC policies? And then run Monte Carlo or other analyses against this? Such modeled events could result in cost increases or decreases and could center around major health health changes (including the onset of chronic conditions), loss of spouse, need to change residence or make expensive living accomodations, need to support a relative or adult child, etc.

    1. Hi, you are bringing up an important point. Mark Warshawsky has moved in the direction of adding these sorts of shocks into Monte Carlo simulations. You can read about it in his book, which is included in the list on the "Books" tab here.

      But more generally, the main point of adding random health shocks is that the "sustainable withdrawal rate" will be even lower to provide greater liquidity for them.

      The way I've tried to deal with this problem is with the efficient frontier approach in which I try to balance between two tradeoffs: meeting as much of one's lifestyle spending goals as possible while also preserving liquid financial assets to cover unexpected expenses.