Over the last year, a group of Bogleheads forum members, including FWF's own LadyGeek, have been helping me to develop a new variable withdrawal method called VPW along with a back-testing spreadsheet to investigate how it would have behaved in the past. Here are a couple of related links:
- Bogleheads discussion thread: http://www.bogleheads.org/forum/viewtop ... 0&t=120430
- Bogleheads Wiki VPW page: http://www.bogleheads.org/wiki/Variable ... withdrawal
The basic idea of VPW is to compute the payment required to deplete a portfolio over N years assuming a fixed growth rate. Then, for every year (1 to N), VPW computes the percentage: payment / portfolio-balance. This percentage is then used to compute the actual portfolio withdrawal in real life, by applying it to the real-life current portfolio balance, every year.
Example:
Code: Select all
N = 30
Growth = 3%
=> Payment= 0.049533
Balance
=======
Year 1: 1.000000
Year 2: 0.978981 (= (1.000000 – 0.049533) * 1.03)
Year 3: 0.957331 (= ( 0.978981 – 0.049533) * 1.03)
…
Year 30: 0.049533
So, we get:
VPW Table
=========
Year 1: 4.95% (= 0.049533 / 1.000000)
Year 2: 5.06% (= 0.049533 / 0.978981 )
…
Year 30: 100%
Of course, it's only a withdrawal method, not a full retirement plan. It can't offer a spending floor (that's the role of CPP, OAS, and pensions). Yet, it seems to me more sensible than the very risky "Safe" Withdrawal Rate (SWR) approach where you compute an initial withdrawal when you retire, then you index it to inflation every year, regardless of market returns, hoping everything will go well (and probably leaving tons of unspent money on the table if things go well, indeed).
So, what do you think?