[You can read the first and second in the series as well.]
So far I’ve just confined myself to looking at whether Prime Harvesting gives you more money (on average, under most circumstances). But there’s more to life than that. After all, on average you’ll make more money by being 100% equities. But that brings more risk than most people are comfortable with. So we need to try to decide if Prime Harvesting puts more money in our pockets by taking on more risk.
10% Stocks: $31,953
60% Stocks: $49,706
100% Stocks: $60,456If we only look at the median withdrawal, it is clear that going 100% stocks is the way to go. So why don’t we do that?
We have some (admittedly not terribly well defined and subject to behavioural biases) notions of risk and we generally feel that a 100% stock portfolio is too risky.
Under this “salary Sharpe ratio” Prime Harvesting looks substantially better than an Annually Rebalanced portfolio.
When looking a portfolio you can use a Sharpe ratio which is a way of calculating “risk-adjusted return”. (A higher Sharpe ratio is better.) A portfolio that is 100% equities has a worse Sharpe ratio than one that is 60/40.
We can try the same thing with income and make a “salary Sharpe ratio”. (Remember, higher is better; I’m using “certainty equivalent withdrawals” and standard deviation of withdrawals to form the ratio.)
10% Stocks: 69,682
60% Stocks: 123,374
100% Stocks: 107,002
Prime Harv.: 133,579
Alt Prime: 127,767Looking just at the annually rebalanced portfolios, this matches our intuition. A portfolio that is 100% stocks will be riskier than a portfolio that is 60% stocks.
Under this “salary Sharpe ratio” Prime Harvesting looks substantially better than an Annually Rebalanced portfolio.
This is a step in the right direction but it is far from perfect. Volatility is not the end-all, be-all of risk. Though with retirement income, it is probably a pretty big factor.
I started this whole series mostly because I learned about Milevsky’s “risk quotient” and wanted to see what things would look like if I applied it to Prime Harvesting. After a large digression, it’s time to go back to “risk quotient”. Will we still Sleep Well At Night given the changing asset allocation of a Prime Harvesting portfolio?
Every year I recompute the risk quotient based on the retiree’s age, asset allocation, and current withdrawal rate. This seems like a pretty good proxy for the kind of ad-hoc risk evaluations that people do. If you are older, have a less volatile portfolio, have higher portfolio returns, or are taking a lower percentage of your portfolio you are going to feel more comfortable.
(Note: I don’t think risk quotient is a perfect measure of this. In particular, people certainly aren’t doing complex mental calculus before deciding they are taking too much risk. But I’m going to make an assumption that the risk quotient is correlated with the personal risk calculations each of us make. That seems like a good starting point, at least.)
What does the risk quotient look like in practice when used this way?
This is what it usually looks like. It starts out high and decreases over time. There are some shocks along the way — recessions or just a few years of low returns. That’s how it almost always works with a variable withdrawal strategy; the variable withdrawals keep the retiree from worrying too much about the portfolio running out of money.
Let’s try using Constant Dollar withdrawals…which quickly shows why you shouldn’t use Constant Dollar withdrawals!
A retiree in 1920 ends up with a portfolio dramatically larger than anything they will ever need, thanks to the bull market of the 20s. (Five years of better than 25% returns for equities! Holy moly! The portfolio tripled in a decade.) By the time the Great Depression rolls around, you can barely see the impact, especially since they are now in their late 70s (and know that the portfolio doesn’t have to last as long).
In contrast, a 1966 retiree sees their portfolio constantly dwindle. Their stress and anxiety mounts until the portfolio eventually runs dry.
So this seems like a promising approach. Let’s add Prime Harvesting into the mix for some comparisons and look at 1920 and 1966 again.
In both cases, the difference is driven largely by the bond allocations. In 1920, Prime Harvesting “over harvests” and you end up with a lot of bonds. In 1966, you end up selling all of your bonds waiting for stocks rebound.
This doesn’t look great for Prime Harvesting. It looks like it has an average “risk quotient” around the same as a 65/35 portfolio. And, if you remember my previous post, we only got about $700–1000 a year more in “constant equivalent withdrawals” in exchange for those sleepless nights.
It seems like a bit of a wash. You get a bit more return for a bit more risk.
(This is a good time to stop and point that McClung would probably object that portfolio volatility (and by extension, Milevsky’s “risk quotient”) is not an appropriate measure of risk for a retiree — who should care much more about income. I think there is a lot of truth to that. However. 1) I think we need to have a firmer theoretical/mathematical/historical/something underpinning for that line of thinking. 2) We know that retirees do lose sleep over the volatility of their portfolios today.)
The risk quotient for Annual Rebalancing doesn’t change very much, regardless of the year we retire in. That’s because three of the four inputs don’t vary: the retiree’s age, the expected returns of the portfolio (constant, since we never change the asset allocation), and the volatility of the portfolio (also constant, for the same reason). And since we are using VPW, the withdrawal percentage each year also doesn’t change between cohorts. So you don’t see large fluctuations.
Prime Harvesting, due to its fluctuating asset allocations, does see variations in risk quotient.
Let’s try graphing withdrawals (under VPW’s variable withdrawal strategy) and risk quotient together:
It is hard to shake the feeling that Prime Harvesting owes a fair chunk of its outperformance to more risk taking. Remember that in 1966 Prime Harvesting actually gave you more money every year, especially once the crisis passed.
But, at least by the risk quotient measure, Annual Rebalancing did a better job of helping you sleep well at night.
It appears to do a little better in return for taking on a little more risk.
In Prime Harvesting’s defense, it isn’t exactly raising the risk quotient to stratospheric levels. We’re usually talking about the difference between a risk quotient of .08 and .09. Still, more risk is more risk.
So where does that leave us?
Prime Harvesting has better withdrawals, though not necessarily by a lot. It handled worst case withdrawals a bit better. With one risk metric — the “salary Shape ratio”, looking at “salary risk” — Prime Harvesting does very well. With another risk metric — the “risk quotient”, looking at “portfolio risk” — it doesn’t do as well.
I don’t think the case is compelling enough for Prime Harvesting to be a “default recommendation” for most retirees.
It appears to do a little better in return for taking on a little more risk.
If you don’t think portfolio volatility is relevant to retirees, however, then you are likely to look more favorably on Prime Harvesting.
