The Prime Harvesting methodology in Living Off Your Money is for use during retirement when you are decumulating assets. But it is pretty natural to wonder, since it outperforms Annual Rebalancing during decumulation…would it (or a variant of it) outperform Annual Rebalancing during accumulation?
I had idly wondered that when I first read the book. When Stefan Hajnoczi emailed me wondering the same thing, it spurred me to look into it. It took me so long that he had done the same thing, so a lot of what is below comes from an email exchange with Stefan.
Triggers during accumulation
The first thing you need to do is tweak Prime Harvesting, since it is built for retirement. Retirement means you have rules like “always spend from bonds” and “sell stocks when they reach 120% of the value they were on the day you retired”. If you’re still accumulating, those rules don’t make sense.
These are the rules Stefan proposed:
- Only buy stocks. Never buy bonds. You will accumulate bonds during the “harvesting” step.
- Harvest your stocks when they have appreciated by (an inflation-adjusted) 20%. This is trickier to calculate since you essentially are tracking the price appreciation of every lot you’ve purchased during accumulation. With computers and spreadsheets it isn’t hard but it is more bookkeeping than Annual Rebalancing by a fair amount.
- Skip the harvesting step if you already have 40% bonds. This limits the drag caused by bonds, especially if you start out your accumulation phase with strong stock returns.
If we just look at final portfolio values, Prime Harvesting Accumulation actually seems to do pretty decently:
There are a few periods when Annual Rebalancing outperforms: 1871–1887 and 1978–1980. But that’s only 3 years in the 20th century. Not too bad. And when it did underperform it wasn’t by very much: the largest underperformance was 2.2% for the 1979 retiree.
More often, though, Prime Harvesting Accumulation outperforms Annual Rebalancing, usually giving you a final portfolio that is about $132,000 higher.
Here’s a zoomed in view of a typical accumulator (from 1950, chosen at random) showing both
That makes it a bit easier to see that Prime Harvesting Accumulation seems to be better, though not necessarily by tremendous amounts. (But since we’re talking about money, there’s no reason to discard small efficiencies.)
In a few cases it does substantially better, though I’m tempted to put that down to “luck” (and the time windows chosen) more than anything else. For instance, for a 1973-accumulator would hit the “20% harvesting trigger” in 1999. So you’ll end up selling $275,000 of equities right before the market crashed. But if we extend the simulation and run it for 40-years (instead of 30-years) that temporary outperformance becomes much smaller.
It is hardly definitive but, at first glance, it looks promising. After all, better returns are better returns.
But what we really care about are better risk-adjusted returns. We only ever buy stocks and rely on the harvesting mechanic to keep us close to 60/40 portfolio.
If it outperforms simply by having a higher stock percentage than a 60/40 Annually Rebalanced portfolio then it is “cheating”. On average, the higher your stock allocation, the better your portfolio will do. Let’s look at that same 1950-accumulator and their bond levels.
The average bond level across all 30 years of accumulation is 38%. It takes a bit under 10 years to get up to the “normal” amount of 40% bonds. And for decent stretches of time, the portfolio has more than 40% bonds.
If the outperformance relative to Annual Rebalancing is due to overweighting stocks, it isn’t immediately obvious from this.
So it looks promising, though I think a better look at the risk-adjusted returns is still needed. But the approach has some intuitive appeal: when your stocks have appreciated a lot, take some money off the table.