I'm afraid that you missed the point. Of course, the 1965/66 cycles go to the ground whatever harvesting technique you use when using constant-withdrawal at 4%. Now pay close attention to what actually happens (check the chart in this post). We start from a highly valued stock market (CAPE around 24). The algorithm makes you sell bonds and the stock part of the portfolio strongly increases, while the stock part of the portfolio can't reach the 120% threshold. Rather luckily, in 1972, the threshold is reached, but the AA remains very high when the oil crisis strikes. As a retiree, I would feel pretty bad about the algorithm at that point. And then as the inflation ravages the US, the AA increases again and stays stuck to to 100%, as the portfolio spirals to the ground. Nobody would follow such recommendation, there is no way.heyyou wrote:MSWR is based on surviving the single worst period, then posters pick that same period to show that it is not a good period for McClung's suggestions. It is not a good period for any retirement. One of McClung's points is that his methods are an improvement over MSWR, allowing the retiree to spend more from the same amount of assets, with more assurance that it will survive a similar worst period, as does VPW.
Now granted, the trajectory of the regular fixed-AA portfolio at 60-40 isn't pretty either, although it does manage the pre-crisis and crisis itself a tiny bit better (and with MUCH LESS regrets). In addition, even if Prime Harvesting delays the portfolio failure a tad for the 1965 cycle, it does not in 1966 (with the set of historical return numbers I used). And how can that help selecting a higher withdrawal rate to start with? One needs a variable withdrawal method to manage those situations, this is clear.
So this all tells me the following:
- the calibration of the entire algorithm on the 120% threshold of the initial (stocks) portfolio is not a very good idea (memory effect)
- one should focus FIRST on selecting a sensible variable method, and THEN ponder about the best harvesting method, instead of the reverse way around
- and then the entire MSWR reasoning is pretty much a moot point.
Am I making sense?