No. of Recommendations: 33
There are usually two goals to be considered when designing a portfolio drawdown strategy.
First, it's nice to get better returns than the average person: a higher average sale price for a higher average income, perhaps an increase in portfolio size.
And second, you want to deal with longevity risk: you want to spend as much as your portfolio as you can before you croak, but never take the risk of outliving your assets: being old is bad, and being broke is bad, but being both is horrible.
I've posted on various possible ways to handle both of these conundrums, since I find both of them interesting subjects. I'll be happy to dig up links to anything that strikes your fancy.
The first issue is best summed up by a simple observation: there are ways to get a better realized price from periodic sales, but at the trade-off of irregular income stream. You can think of it as a tuning knob: more irregular liquidations and higher long run income, versus steadier and lower.
If you can handle irregularity in return for higher returns/income, there are schemes that I can suggest.
One method I mentioned assumes nothing about the valuation of the stock, but merely sells a bit each time there is a new high price. However, as the time since the most recent sale lengthens, it "decays" the target price representing the previous high. This is an example: green dots represent days you'd sell some stock.
http://stonewellfunds.com/SWRpri075.jpg The blue line is of course the stock price, and the yellow line is the target line for a new sale: high to date, falling gradually. Note that the average level of a green dot (sale) in any stretch of 2-3 years is a fair bit higher than the average level of the blue line (stock price), maybe 3% or so.
At the other extreme is steady selling. My own will suggests that my accountant and spouse arrange the following: put half the portfolio into BRK stock and half into a fund I've chosen. Each quarter, sell 1.4% of however many shares of each one are currently remaining. (any dividends that have arrived during the quarter can be considered like shares that have already been sold, reducing the sale size). The income is irregular only to the extent that the price varies. By construction, the portfolio will have the same real value 20 years later if the investment have an underlying return of inflation + 5.80%/year. If returns are better than that, the income will rise (irregularly) over time. But the scheme's success does not depend too heavily on that number...even if the portfolio manages only inflation + 4.0%, the portfolio will still have 71% of the starting real value at year 20.
A good strategy really should adapt to the progress in the value of the underlying asset. Starting to deal with the second goal at the top, the limit on how much you can withdraw from any portfolio is ultimately capped by the value generation of that portfolio. So, if you can estimate the value, you can safely liquidate the amount by which it has risen in real terms since you started your sale program. If you don't exceed that, your income may vary and could fall to zero, but you're guaranteed never to run your portfolio down to zero by selling too much. For this strategy, I have suggested something like this: each quarter, estimate the value of a Berkshire portfolio based on the 16 most recent values of book per share, adjusted for inflation, using a "weighted moving average" (WMA). That just means that the most recent figure gets a weight of 16, the second-newest gets a weight of 15, and so on down to a weight of 1 for the oldest figure. The smoothing is calculated to be just enough to keep the estimated value rising (though more slowly) through dips in book value due to bear markets, while adapting automatically to medium to long term changes in the rate of growth of the value of the firm. Then, just calculate how much this metric of value has risen since the last time you checked, and sell the difference: to the extent that your value estimate tracks the stock's value over time with decent accuracy, your portfolio will by construction always have the same real (after inflation) value. Possible elaboration 1: don't sell the amount recommended, sell 90% of that. This adds a bit of conservatism, and should lead to the real value and real income rising a little bit over time. Another elaboration 2: I have a lot of confidence in Berkshire. I think you could safely put a floor on the withdrawal amount: if the sum of the last four calculated liquidations (three actual plus current recommended) is less than 4% of your original starting portfolio value, go ahead and sell a bit more so you're always getting a minimum 4% real withdrawal rate in any rolling year. With those two elaborations, a $1 million portfolio started in 2000 would have a real value of $1.20 today, a rate of return of inflation + 0.79%/year, after having paid out an average of $79126/year in real terms in any rolling year, ranging from $40000 to $137590.
For longevity risk, the only really sensible solution is a pooled scheme. The chances are by definition low that you'll live to an unusually ripe old age, so it doesn't make sense for every individual investor to hold back a large fraction of their funds against that remote possibility. Better for everybody to put aside a little into a common pot, and use that pot to support the few people who end up needing that income. So I usually recommend a two part solution. Estimate your life expectancy age and add a bit. Buy a deferred annuity (or tontine if you can find one!) to handle your income after that date till your death. Depending on your age, this might be 5-15% of your capital at a guess. Then, spend the rest of your portfolio in a roughly straight line from now till then. (each year with N years remaining on the straight line, go ahead and spend 1/N of the portfolio). Just as your personal portfolio runs out, your pooled scheme kicks in and supports you for life. Annuities are a total ripoff in many ways, but they do solve a real problem, and most of the disadvantages (hugely negative expected internal rate of return) don't apply if you get one when you're already quite old. Possible refinements: don't buy a very-deferred annuity. Instead, put that money into inflation protected bonds maturing when you need that income, or not too long before that. When the bonds mature, buy the annuity. An immediate annuity producing $X income starting age 85 is much cheaper if bought at age 85 (after 20 years of mildly positive real bond returns) than a deferred annuity purchased at age 65, plus you skip 20 years of credit risk and inflation from the insurer and leave a much bigger estate if you die sooner. The only disadvantage is it's a little harder to calculate precisely how much to set aside.
A couple of random annuity data points: a male Canadian age 80 can get an immediate single-life non-inflation-adjusted annuity paying 9630/year these days. Probably more, that's just the first number I found. Breakeven not counting inflation is at age 90.4, and life expectancy is about 89, so the loss is not large. The advantage is that it covers you for life, as there's a 10% chance of reaching age 96+, and 2.7% of reaching 100+. It is usually better to use extra money to buy a larger non-inflation-protected annuity than to buy the more expensive partly-inflation-protected one, as they tend to overcharge for inflation protection and in any case it is usually only partial. You can also use a small portion of the annuity budget to buy a second deferred non-inflation-protected annuity, to top up the real income starting after some years: also cheaper than an "inflation protected" product.
I can probably post links for more information on any of these that might be of interest. What can I say, I'm a mortality geek.
Jim
