Thursday, November 17, 2011

Length of Retirement and Safe Withdrawal Rates

Brad asked a question in the comments of my last blog entry:

Wade - thanks for this post. I know it's still a work in progress but it's the first chart I've seen that shows MSWR for a 15 year retirement. I'm a Financial Planner that focuses on retirement income planning for my clients. Most research (Bengen, Kitces, etc) shows 30 year SWR's. But I recently had a client in his late 70's come to me to discuss retirement income. Seems like a 4% SWR would be inappropriate for someone with what could be a less-than-30 year lifespan. Michael Kitces and I emailed about this and he mentioned that even he hadn't done (or seen) the research on 15 - 20 year SWR's. Which brings me to my point: as a client ages and their retirement timeline moves from 30 to 20 to 10 years, the SWR should (I think) be raised to counter the effect of a declining lifespan. I, for one, continue to focus on a 4.5% - 5% SWR (I use Bengen's research plus Cuts/Freezes/Raises advocated by Guyton & Klinger) even when a client is in his/her 70's/80's. But, that doesn't "seem" right. Anyhow, thanks for letting me spill my thoughts and thanks for your research on this subject.

It's an important question!

The Trinity study does show portfolio success rates for retirement durations between 15 and 30 years.  But let me offer two other looks at it.  The first figure here shows William Bengen's SAFEMAX (the worst-case sustainable withdrawal rate from history since 1926 for a 60/40 portfolio of large-cap stocks and intermediate term government bonds) for retirement durations up to 40 years.  Indeed, shorter retirement durations allow for higher safe withdrawal rates.  For instance, for a 10-year duration, the lowest ever sustainable withdrawal rate in inflation-adjusted terms was 8%.

The next figure is based on Monte Carlo computer simulations of the same historical data as used in the previous figure.  It shows the "safe withdrawal rate" as defined by that which has a 10% chance of failure for different asset allocations and different retirement durations. Again, you can see that someone planning for either shorter or longer retirement durations should not necessarily be basing their decisions on the default 4% rule, as that comes from a 30-year retirement duration.

7 comments:

  1. Thank you for this wonderful post.It makes sense.Good job!

    ReplyDelete
  2. Wade,

    This is really interesting data- I would like to see the curves in more detail. I don't suppose you could add a link to an excel spreadsheet with the data?

    -Rick Francis

    ReplyDelete
  3. Rick, thanks for the question. Sorry for my delay.

    For the first figure, here are the lengths of the retirement periods (1-40 years) along with the maximum sustainable withdrawal rates:

    1 100
    2 43.50
    3 27.12
    4 20.07
    5 16.34
    6 13.55
    7 11.51
    8 10
    9 8.92
    10 7.98
    11 7.34
    12 6.84
    13 6.43
    14 6.14
    15 5.90
    16 5.68
    17 5.44
    18 5.25
    19 5.09
    20 4.94
    21 4.79
    22 4.67
    23 4.55
    24 4.45
    25 4.36
    26 4.28
    27 4.21
    28 4.15
    29 4.09
    30 4.03
    31 3.99
    32 3.95
    33 3.92
    34 3.89
    35 3.87
    36 3.84
    37 3.82
    38 3.79
    39 3.77
    40 3.74

    ReplyDelete
  4. For the second figure, how about I provide the optimal points, which are connected by the thick black line.

    Here we have retirement duration, optimal stock allocation, and the maximum withdrawal rate... based on Monte Carlo simulations and allowing for a 10% chance of failure:

    10 28.06 10.11
    15 37.05 7.09
    20 38.33 5.65
    25 46.04 4.83
    30 44.76 4.32
    35 46.04 3.96
    40.00 57.60 3.71

    ReplyDelete
  5. sorry, these Comments are sucking the spaces out. Maybe you can copy and past those into a spreadsheet with space as the delimiter to be able to see them better

    ReplyDelete
  6. All these withdrawal rates must take into account your Top Ten Reasons?
    In other words it all has to be in no taxable accounts etc.?

    ReplyDelete
    Replies
    1. Yes, you are right. These are the simple numbers before accounting for the 10 issues.

      Delete