yankees60 wrote: ↑Sat Aug 12, 2023 8:32 pmAlso, don't we really only have a shade over 3 totally independent 30 year cycles? 3 is a tiny, tiny, tiny, tiny sample.
In more extreme cases I extend mine out to around 11. US (from 1871), UK (from 1896), Japan (since 1972). Spans a broader range of events such as decline of the Pound as the predominant international trade settlement currency etc. But does exclude total collapse (Japan and WW2 for instance). Broadly the US was a right tail/good-great case, UK was more middle road (more reflective of global middle/average).
With heirs I also focus more on PWR, looking to leave a legacy of around the inflation adjusted start date value. 3.33% 30 year SWR return of inflation adjusted money via yearly instalments, ending with wealth preserved.
Better in that respect for me is a Talmud thirds each land (homes), stocks, gold asset allocation. Where imputed rent is also considered as part of that, historically 4% so 1.33% proportioned, leaving 2% of total wealth as disposable income. I also set 'inflation' to be 50/50 CPI and average of house/stock/gold price only. Historically house/stock/gold price only inflation was > just CPI inflation alone. i.e. anticipate 2x total return, 1x via SWR/spending (so CPI inflation rate), 1x left at end (that might be 'spend' on buying equal amounts of land, stock, gold).
In regards to rebalancing, diversify across both rebalanced and non-rebalanced. Home(s) value left non-rebalanced (illiquid). For currency diversification, UK home, US stock, global gold.
Improvements in technology/efficiencies have enabled CPI to relatively lag other assets inflation rates. Fields full of manual workers replaced by a single individual and a big machine that can do the same work in a fraction of the time. Perhaps such efficiencies are near peaked, so in future perhaps CPI and other assets price inflation rates might be more inclined to align. Or maybe not.
A nice feature is that portfolio total return growth versus price only inflation rates tend to more align, portfolio volatility is reduced - that is inclined to yield a better result than if variance was higher. Also reducing SWR even a little can make a big difference to outcome. Having part of SWR as liability matched imputed rent and the volatility in rental yields is irrelevant when you are in effect both landlord and tenant.
The negative side is that of potential regret of looking back and seeing how much more one might have had, had they only been more aggressive in their asset allocation. Maybe 3x or more wealth when having ended with 'just' 1x. But equally to target that 3x type objective involves taking on more risk, so one risks outcome for self in favor of potential benefit for heirs, who may just blow it all anyway. Better IMO to target a modest/reasonable legacy and having lower risk to oneself instead of higher risk to oneself.
Should stocks become no longer viable/possible, they'd be replaced with art. Thirds each land, art, gold ... old-style asset allocation. Each might broadly offset inflation, but in a volatile manner. With diversification across all three the tendency is towards closer alignment with inflation, as one lags so another might lead, combined smoother average. With stocks you have both price appreciation and dividends, as with a house you have price appreciation and dividends. If stocks were swapped out for art you give up the dividend benefit (as do you when holding gold). Broadly imputed rent and dividends tend to compare (as do house and stock prices). So land/stock/gold when you assume imputed = dividends is somewhat 67/33 stock/bond like, a third less 'dividends' than all-stock. The improved (reduction) in volatility however narrows down that gap, less variance, better compounding.
Which fundamentally is my PP is just the stock/gold elements. No bonds, land instead. Or could be considered to be 67/33 equities/gold. With a 3% SWR applied to that stock/gold liquid assets value. Given the lower stock weighting I prefer small cap value instead of TSM. As yet further testing you can use the likes of
Monte-Carlo simulations of data, which might be considered as adding a 12th non-overlap data set