There is a subtle difference, though it has a big implication. The difference is that you don’t independently choose a withdrawal rate and a wealth accumulation goal, but just directly save enough to finance your desired retirement expenses. It leads to the differences between the black curve and the blue curve in Figure 5.

From page 3:

“I am suggesting that the following retirement planning process, which is one which isolates the accumulation and decumulation phases, is not appropriate.”

The author goes on to summarise current financial planning practice to be (the steps listed below have been reworded so as to read from the third person);

Step 1: Estimate the withdrawals needed from financial assets to pay for planned retirement expenses after accounting for Social Security, defined-pension benefits, and other income sources. Planned retirement expenses are defined as a replacement rate (RR) from pre-retirement salary.

Step 2: Decide on a comfortable withdrawal rate (WR) shown to be sufficiently capable in the historical data.

Step 3: Determine the wealth accumulation (W) retirement goal, defined as W = RR / WR.

Step 4: Determine the savings rate (SR) needed during working years to achieve this wealth accumulation goal.

Donald suggests replacing Steps 2 to 4 of the above process with the following:

Step 2: Decide on a comfortable savings rate (SR) based on what has been shown to be sufficiently capable of financing desired retirement expenditures in the historical data.

On page 4, Donald then goes on to define what a sufficiently cable savings rate should be.

“A particular savings rate was successful if it provided enough wealth at retirement to sustain 30 years of withdrawals without having the account balance fall below zero. Actual wealth accumulations and withdrawal rates may vary substantially for different retirees.”

This statement is simply a restatement of Steps 2 to 4. Put another way, this is what Steps 2 to 4 do leading the reader to wonder what, exactly, is new with this approach.

Note that investment income used to pay rent is taxed (and may trigger taxation of Social Security benefits). Yet if you “invest” in your home, the “income” of low-cost housing is entirely tax-free and does not count towards your AGI.

Yet another example of how income taxes create truly perverse incentives. The homeowner is likely already wealthier than the renter, so the renter is rewarded by paying higher taxes in retirement?!?

Any views on the investment outlook should be relative to the balanced portfolio and considered a separate problem (though not a separate optimization). Further, the investor could also dynamically adjust the risk aversion coefficient of the balanced portfolio (though not in the reverse optimization process) based on market conditions.

To the extent that he would prefer a constant savings contribution to take advantage of mean-reversion, then he is not necessarily passive. He does have an opinion on the return generating process. I would prefer to keep active and passive decisions separate. For instance, you could assume that the returns over your original optimization time horizon will be distributed as your inputs into the optimization (which could be influenced by views) and then normally distributed according to the original covariance and reverse optimized mean until say 30 years after death. You could then set your savings rate such that the probability that your wealth 30 years after your death is less than 50% of your final salary is less than 5%.

This is obviously more than any individual investor could do, but it wouldn’t be so hard to program into something a financial planner could use.

Whether you invest in the stock market, in bonds, in real estate, or in “hard commodities”, be certain that the free cash flow generated is sufficient for your needs. Anything beyond that is speculative.

Doug Short occasionally posts a “Total Return Roller Coaster” chart on his website that looks at S&P 500 rolling 10, 20, and 30 year total returns: http://www.dshort.com/articles/SP-Compos ite-annualized-total-return-roller-coast er.html . Since few of us have a large lump sum that can just sit over 30 years without inputs or outputs, I think the 10 and 20 year charts are the most valuable for evauating investing and withdrawal stragies for the typical person.

A couple of things leap off the page of these charts:

1. Subzero annualized real total returns over 10 year periods can persist for up to a decade – it only recently went below zero in 2008; and

2. Once the 20 year real total return drops to below 5% annualized, it invariably goes to near zero within a few years indicating a long period of poor returns. It has only recently dipped below 5% in 2008-2009.

Both of these charts indicate to me that we are likely to have another significant market decline over the next few of years. After that we are likely to begin a long upward trend that likely could generate real total returns of 8% or more per year for a decade or more.

Similarly, interest rates are very low now (BND yield = 3.5%) so income from bonds over the next decade should be assumed to be very low with the potential for capital losses over that period.

As such, my savings strategy is simple: I need to be saving 20%-30% of my current income for the next decade as I cannot count on the stock or bond markets to deliver decent total returns to build up my nest egg during that period. However, this makes it a good period for long-term investing as I expect that after retiring in 10-15 years, I will likely be able to pull money out at greater than Bengen’s 4% safe withdrawal rate (probably at 5%-6%) if the equity valuations are relatively low (less than Shiller’s PE10 of 16) at that time.

At retiremement, the expected total return from equities based on their PE10 valuation at that time plus the interest rate on a total bond fund is a good starting point for determining whether or not a safe withdrawal rate can be bumped up to 5% or 6%.

The current PE10 of about 23 ( http://www.dshort.com/articles/SP-Compos ite-pe-ratios.html) and the BND yield of 3.5% tell me that the 8% returns assumed by many pension fund managers today are ludicrous at best and criminally fraudulent at worst. Just another example of the governing class of people gaming the system to buy themselves another handful of years raking money off the top before their bubble world bursts.

At least I can make a decent stab at understanding how much my 401k/IRA is really worth looking at it in a multi-decadal way. My other source of valuation is to pay attention to what insurance companies are willing to payout in annuitized payments. They seem to be the only part of the financial sector living in the same world that the average person inhabits.

And what is the payout rate on that? Insurance companies demand a large profit on the annuities that they offer, and an additional profit on top of that if they are taking any risk at all (such as promising inflation-adjusted payouts).

The least risky strategy is to give your life savings to charity and join a convent. You are 100% guaranteed to die poor if you do that. Buying COLA annuities from insurance companies is barely any better than that.

Let’s see… 30% saved * 30 years of saving / 30 years of retirement = 30% replacement. I can’t live on 30% replacement, can you?

Real returns may “only” be 2% to 4% over a lifetime of savings and investing, but that 2% makes all the difference!