Utilizing CAPE for a safe retirement withdrawal rate

Target-date funds have risen significantly, thanks in part to the Pension Protection Act of 2006, which enabled automatic enrollment.

Target-date funds, for the most part, follow the old investment advisor adage that a person’s stock weight should be around 110 minus his or her age.

While this fits the common-sense notion that, all things being equal, a person’s stock weight should decrease as they age, it has several flaws.

Bottom line a risk-management framework based on a common-sense notion of risk: not having enough money in retirement is a more reasonable approach than the glide path that is part of the target date solution.

The objective is simple: enhance the likelihood that we’ll have adequate finances in retirement.

Unfortunately, the typical solutions (strategic asset allocation, target-date funds) fail to take into account the fact that expected returns change with time.

First of all, even if returns are evenly distributed, the wealth created by those returns is not.

Normally distributed returns aka a bell curve which is what strategic asset allocation and target-date funds assume is not relevant though. Relevant is the following chart which shows the distribution of ending wealth after investing $1 for 40 years in an asset with the normal return distribution.

The right tail of the distribution dominates expected, or mean values.

It doesn’t matter what your target date allocations were if you were fortunate enough to live and save during a period when asset returns were high.

The left-hand side of the distribution is more important – those times when asset markets were unkind and returns were scarce.

Those are the circumstances in which lifetime ruin, i.e. running out of money in retirement, is a distinct possibility.

The best method to create retirement portfolios is to focus on how much wealth is required and when it is required, rather than on maximizing projected wealth, but on minimizing the expected wealth shortfall from what is required in retirement.

Systematic Withdrawals

We are all well aware of the pitfalls of a tradition systematic withdrawal approach to retirement income
especially in light of what we know about sequence return risk. What if instead of simply assuming we
earn 10% on stocks, 5% on bonds, and 2.5% on cash equivalents and withdrawing 4% we utilize forward
looking return assumptions based on fundamentals (stocks) and yields (bonds, cash)?

A safe relative withdrawal rate utilizing forward looking return assumptions

Is this such a novel idea? In Ed Easterling’s paper, “Markowitz Misunderstood,” Ed brings home the fact
that there are two requirements for utilizing a modern portfolio theory approach to finding an
appropriate asset allocation mix:

  • An assumption about future returns
  • A correlation matrix

Ed also questions whether the financial services industry got it wrong when they simply choose to use
return assumptions based on the average annualized return over 85+ years of data on the S&P 500. His
point is well taken in light of what happened to strategic asset allocation modeling during the ’00-’02
and ‘08/’09 stock market meltdowns.


Regardless, what if we were to use what we know about Shiller’s work and the cyclically adjusted price
to earnings ratio (CAPE) on stocks represented by the S&P 500 and the internet to access yield data on
fixed income expectations.


If we look at the Shiller research we then find that CAPE rankings and subsequent 10-year average
returns have the following relationships.

Therefore, if we have access to the internet we can easily find what the Shiller CAPE value is (to
determine our return expectation for stocks) and what the yield on a 10-year bond and the 90-day T-Bill
yield is to determine our fixed income return assumptions.

Safe Withdrawal Rates


In 1952, the College Retirement Equities Fund (CREF) was started to compliment TIAA, therefore
providing teacher’s a combined balanced retirement stream for their non-discretionary spending (TIAA
fixed annuity} and discretionary spending (CREF variable annuity). Both TIAA and CREF rest on an
assumed investment return (AIR). In the case of CREF, the AIR was 4% (not sure what it is today). This
AIR sat on a simple notion that if you want to sustain your asset base over time while drawing income
from it then you better utilize a realistic return assumption on the underlying vehicle (stock indexation)
supporting that assumption. I left TIAA-CREF in 1995 and was not there in ’52 when the specs were
drawn but given the idea at the time when I worked there that stocks “were a good hedge against
inflation” I’m guessing that the CREF AIR represented a 3% inflation assumption, a 3% dividend yield
assumption, and a 4% capital appreciation assumption which all total 10%. Pretty much the same figure
the Ibbotson Associates (Morningstar) folks used to assume the “long-run” returns on large cap US stocks.

So where does this leave us?

Right back at the traditional 4% safe withdrawal rate (SWR)?

How about a more dynamic conclusion…


What if we were to assume that the 4% SWR represents 40% of our long-term 10% return assumptions
on stocks and made the assumption that to be prudent we would determine our safe withdrawal rate
using a 40% rule of thumb based on our long-term return assumptions.


So for example, in December 1999, the Shiller CAPE was at 44.20 (an all-time high) which translates into
an AIR of 1%, 10 year Treasuries at the time were yielding 6.28%, therefore depending on your mix,
below are the appropriate safe withdrawal rates.

On the flip side, what if we retired in July of 1982 (the start of the greatest secular bull market in the
history of recorded US stock prices) when the CAPE was 6.62 and 10 year treasuries were yielding
13.95%? Here would be our safe withdrawal rates: