In this post, I explain in details why I removed the
portfolio size warning in version 2.8 of the worksheet.
A hypothetical 50 years old retiree has a $1,000,000 balanced portfolio (60/40 stocks/bonds) and anticipates to get a $50,000 (annual) pension indexed to inflation (such as Social Security) starting at age 70. There's a 20 year gap between retirement and the start of pension payments.
Assuming some cash investment (like I-Bonds, CDs, etc.) could deliver approximately 0%
real per year for the next 20 years, the retiree could put aside
$50,000 into a savings account for this year and put
($50,000 X 19) = $950,000 into cash investments for the remaining 19 gap years. The total cost of the pension bridge would be
($50,000 + $950,000) = $1,000,000. The retiree would be burning the entire portfolio to bridge the future pension and be left with no portfolio at age 70. The retiree would reliably get $50,000 per year (adjusted to inflation) for life starting at age 50.
The first version (1.x) of the VPW worksheet supported bridging future pensions using the portfolio. The
VPW Table percentage for a 20-year depletion schedule and a 60/40 portfolio is
6.9%. As a consequence, the worksheet calculated an approximate
($50,000 / 6.9%) = $720,290 cost for the pension bridge, leaving
$279,710 for a normal withdrawal using the 4.3% VPW Table percentage at age 50, that's
($279,710 X 4.3%) = $12,037, resulting into a total income of
$62,037 in the first year of retirement.
But, a portfolio fluctuates. If the balanced portfolio was to lose
-30% (due to stocks losing half of their value) just before making the first withdrawal, it would shrink to $700,000, which is less than the full cost of the (portfolio) pension bridge. To protect the portfolio from premature (before age 70) depletion, version 1.x of the worksheet reduced bridge withdrawals in proportion of bridge funding ratio. In other words, the bridge withdrawal would have been
(($700,000 / $720,290) X $50,000) = $48,592, instead of $50,000, after the loss. Nothing would have been left for an additional normal withdrawal. If the portfolio doesn't recover sufficiently, it gets depleted at age 70, like the cash bridge.
Version 2.x of the VPW worksheet added a
50% of portfolio cap on bridge funding to preserve lifelong liquidity. In other words, the worksheet assumes that only 50% of the current portfolio is available for pension bridges. If this is insufficient, bridge withdrawals are reduced in proportion of bridge funding ratio.
Getting back to our example, this means that only $500,000 is available for bridging the pension. It's less than the estimated $720,290 cost. Consequently, the worksheet reduces the bridge withdrawal to
(($500,000 / $720,290) X $50,000) = $34,708 leaving $500,000 for the normal VPW withdrawal. That's
($500,000 X 4.3%) = $21,517 for a total retirement income of
($34.708 + $21,517) = $56,225.
But, as I previously wrote, a portfolio fluctuates. A drop to $700,000 just before withdrawal would change the picture. The cap would become $350,000, resulting into a
($350,000 / $720,290) = $24,296 bridge withdrawal. The normal withdrawal would be
($350,000 X 4.3%) = $15,062 for a total retirement income of
($24.296 + $15,062) = $39,358 after the loss. That's 20% less than $50,000, but liquidity is being preserved and hopefully, if the portfolio recovers, total retirement income increases back above $50,000.
When I designed version 2.x, I considered the two limits (assuming constant returns in a model):
- Full bridging: when the pension bridge costs 100% of the portfolio, income is level all retirement long, but it leaves the retiree without a portfolio when the pension starts.
- No bridge: no money is spent on the bridge. Until the pension starts, only normal VPW withdrawals are made. When the pension starts, total retirement income increases by the annual pension amount.
None of these two limit possibilities was appealing to me. So, I chose the middle position: use no more than half the portfolio for bridge withdrawals and the rest for normal withdrawals. That's where the 50% comes from.
What I hadn't analyzed, until now, is the long-term impact of this choice. The worksheet has no memory of the past. It only uses current information to make its suggestions. As a consequence, I had no confidence that the 50% of portfolio cap would behave acceptably with large deficits over long periods. I worried that maybe the residual portfolio would possibly shrink to almost nothing. That's why a "
Portfolio is too small..." warning was shown when the estimated cost of bridges was bigger than 50% of the portfolio.
By analyzing it, I've discovered that the 50% of portfolio cap is significantly more robust than I anticipated. Using a constant return projection, the
($500,000 / 720,290 - 1) = -31% deficit shrinks to
0% at age 64 while total retirement income increases from
$56,225 to
$65,116, where it stays for the remaining of retirement. At age 70, the residual portfolio is
$279,298 before withdrawal.
Repeating the calculations after loss, assuming no recovery (other than the normal 60/40 stocks/bonds growth trend), the
($350,000 / 720,290 - 1) = -51% deficit shrinks to
0% at age 67 while total retirement income increases from
$39,358 to
$57,662, where it stays for the remaining of retirement. At age 70, the residual portfolio is
$141,571 before withdrawal.
If I allow the portfolio to gradually recover from its
-30% loss over 20 years (in other words, after 20 years, the total cumulative return is identical to not having had the loss), total retirement income increases from
$39,358 to
$50,743 at age 60, continuing to
$64,391 at age 70 (
-1% less than the no-loss scenario), where it stays for the remaining of retirement. At age 70, the residual portfolio is
$265,906 (
-5% less than the no-loss scenario) before withdrawal.
Using various synthetic scenarios, with small, medium, big, and huge deficits, over short, medium, and long gaps, testing unfavorable return sequences (like losses before first withdrawals), I was unable to find a case where the 50% cap wouldn't have delivered an acceptable outcome, given the analyzed situation. I am positively surprised by the robustness of the 50% cap for bridges.
An interesting case is an age 50 retirement with a $1,000,000 balanced portfolio, but with a $25,000 pension at age 70. The cost of a cash bridge would be
(20 X $25,000) = $500,000, resulting into a total retirement income of
$46,517 for the no-loss scenario and of
$40,062 after loss. Using a portfolio bridge (with the 50% cap) results into a total income of
$52,636 for the no-loss scenario and of
$39,358 after loss (gradually increasing to $39,664 at age 52 and staying there). In other words, when using the portfolio for bridging the 20 year gap, instead of a cash bridge, there's a significant (+13%) bonus for the no-loss scenario and practically no difference (
-1%) after loss when assuming that the
-30% loss isn't recovered during gap years.
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)