Bucket vs Constant Weighted

[1nvest]

A individual about to retire might opt for one of two options. Being 60 they might protect a 25 years lifetime and for longevity/heirs might look to preserve some capital. Drop 62.5% into 'cash' and draw that down over the 25 years, 2.5%/year. The rest (37.5%) is invested into stocks for growth. Assuming the growth is 4% real (after inflation) then that 37.5 grows to 100 over 25 years i.e. they end up with the same inflation adjusted capital available at the end of 25 years. Broadly that starts with 37.5% in stock, ends with 100% in stock (as all of 'bonds' (cash) is spent over the 25 years, so averages 69% stock over the total period - let's round that to 70%. So the choice are to either use the bucket approach for a 2.5%/year income drawn from cash reserves, or yearly rebalance to the broader 70/30 average stock/cash level and draw a 2.5% income.

On the one hand we have interest rates at very low levels, valuations are relatively high. The bucket approach starts with relatively little in stocks and sees that expand over time, a form of cost averaging more into stock exposure over time. If stocks do relatively poorly typically cost averaging in across that tends to do OK. If stocks do well typically its better to have lumped in at the start. So on a valuations measure (low present interest rates/high prices) the bucket approach would seem the current better choice. However cash is also a poor investment at present. Ideally for the bucket style 62.5% being dropped into 'cash' for 2.5% x 25 years of drawdown (income) a inflation pacing (or better still beating) 'cash' deposit would be the more ideal, Index Linked Gilts however at present however are paying negative real returns. Which on that basis potentially makes the constant weighted choice the more appropriate. A close call, so in such cases opting for 50/50 of both, diversifying, is typically can be a good choice. However the bucket approach for someone aged 60 provides better assurance of income. Whilst it presently costs more to buy £10K/whatever of inflation adjusted income in 1,2,3,....25 years time at least you know exactly how much more the cost is from the offset ('fixed income'). That assurance of income is perhaps better than the size/amount that might be available for longevity/heirs.

Perhaps a 60 year old that will only have a 9K/year state pension when they turn 67, that has enough to put aside 63K as a 9K/year substitute for that state pension for ages 60 to 66, and that is or is planning to downsize their city suburb £650K value home for a rural £300K home, along with a additional 100K of savings/investments will have a £450K initial pot of money that at a 2.5% withdrawal rate provides a additional 11K/year income (that combined with the 9K state pension = 20K/year disposable 'pension' - which for them is perhaps 'good enough'). So loads 37.5% into a stock accumulation/growth pot, dropping the rest into inflation bonds.

At recent -2% real type prices for index linked gilts/inflation bonds, compound that year by year and to buy 25 of income across 25 years costs 32 of present day money. A average 'loss' of a 0.78 factor. So on that basis/measure 10K of present day money is only buying 7.8K of future inflation adjusted income. So instead of 11K/year as above that's 11K x 0.78 = 8.6K/year on top of the 9K/year state pension amount, 17.6K/year combined instead of 20K/year had inflation bonds been priced to pay 0% real. Maybe OK still, maybe not, in which case the £60K total 25 year shortfall (25 years x 2.4K/year difference) might have to be otherwise found or achieved. Or perhaps opt for the constant weighted approach and hope for the best. Of the two however the greater certainty of the bucket approach has appeal for those that prefer greater certainty.

Personally I'm more in the stock/gold constant weight camp. 50/50 stock/gold as the 'growth' holding as whilst in better times that lags all-stock, in less good times its more resilient, more inclined to yield a 4% real growth rate. So a bucket approach would start with 17.5% in each of stock and gold, remainder 62.5% in 'cash', that over the 25 years transitions to being 50/50 stock/gold (all of cash spent). Which in constant weighting terms is near-as 33.3% in each of stock, gold and cash. Given low inflation bond yields at present holding the constant weighted choice has perhaps the better potential reward to match/exceed inflation. So pretty much a third each in stock/gold/cash periodically rebalanced with a view to a 2.5% income production rate.

A nice feature with SWR is that typically SWR + remainder average real gains tends to be better than just real gains alone. SWR is form of cost averaging out, which tends to be better than letting things ride. Which can bolster the actual SWR by a percent or so. Maybe a 3% real gain from a accumulation asset allocation might see a 2% SWR + 2% real gain by comparison.

Check out this link, US data, but a very neat/useful tool for visualisation. I selected 1985 as the start year as in my books that was a relatively middle road (broad average) year. I've set a 3.5% SWR i.e. 3.5% of the initial portfolio value and where that capital value/amount is increased by inflation as the amounts drawn in subsequent years - so a nice inflation pacing income over time (none of the HYP as a annuity type approach where dividends can vary massively from year to year). 1% higher than the 2.5% figure we used above in reflection of SWR + real tending to be > real alone (0% SWR). That additional benefit reduces the loading required, potentially achieving the same amount of £ income being provided from less capital, or more income being provided from the same amount of capital.

Give that link/site a try, click the inflation adjusted boxes ...etc. vary the start/end dates etc. Note also that I used mid cap for the stock holdings rather than the main/large cap. I like mid caps as they feed both in and out of the top and bottom and tend to be more equal weighted like rather than cap weighted like. That removes a significant risk IMO such as in Japan, where some stocks might otherwise rise to being massive, then falter and perhaps halve or more, but still remain within the index as their size/scale is so large they still dominate the index. A broader index carrying such stocks is a drag factor, one that mid-cap indexes tended to eject/replace. But if you prefer otherwise then change the stock allocation to total stock market ... or whatever.

As a alternative to a annuity, or using HYP as a annuity replacement, the above is yet another viable alternative IMO.

[mark88man]

Interesting approach, but in this para

At recent -2% real type prices for index linked gilts/inflation bonds, compound that year by year and to buy 25 of income across 25 years costs 32 of present day money. A average 'loss' of a 0.78 factor. So on that basis/measure 10K of present day money is only buying 7.8K of future inflation adjusted income. So instead of 11K/year as above that's 11K x 0.78 = 8.6K/year on top of the 9K/year state pension amount, 17.6K/year combined instead of 20K/year had inflation bonds been priced to pay 0% real. Maybe OK still, maybe not, in which case the £60K total 25 year shortfall (25 years x 2.4K/year difference) might have to be otherwise found or achieved. Or perhaps opt for the constant weighted approach and hope for the best. Of the two however the greater certainty of the bucket approach has appeal for those that prefer greater certainty.


I think that should be a loss factor of 0.68 not 0.78. Changes the figures somewhat but not the sentiment

[1nvest]

I think that should be a loss factor of 0.68 not 0.78. Changes the figures somewhat but not the sentiment

Typed/posted that message whilst bored (waiting), using a uncomfortable device, so initially on seeing your post I thought I'd made a mistake, however on rechecking if I discount 10K/year at 2%/year rate I get a figure of a average 1.28 being required to buy 1.0 future inflation adjusted value out to a 24 year timeframe (first years income being immediately available). And 1 / 1.28 = 0.78.

Just checked US data and for all 25 years starting from 1972 the average real gain growth saw 37.5 stock/gold initial allocation (62.5 put aside for 2.5%/year income) grow to a inflation adjusted 130 (30% more than at the start, which is 1% annualised). Two bad cases of 1980 - 2004 and 1981 to 2005 that both ended at around 85% of the inflation adjusted total start date portfolio value. And in both cases a couple year more saw 100% of the inflation adjusted start date value. Another factor is that at times during those historic years you could lock into inflation bonds/index linked gilts priced to perhaps +3% real rates.

Locking for 25 years at current rates is a fundamental reason why I believe the constant weighted approach to be more appropriate at present valuations. 33% in each of stock/gold/cash perhaps a 3% SWR (PWR) and where cash could be a three year ladder/whatever, that retains the option to move over to a bucket style if/when real yields on inflation bonds move back into positive real yield territory.

[mark88man]

I saw 32% Then I read that you implied this was a 0.78 ratio - so my simple thought I was a typo as 100%-32% is 68% or 0.68. I see now it is the ration of 25 and 32. don't know where I got the % from - As long as you are happy, and as I don't want to derail a careful post, I am happy for my comments to be moderated, by which I mean deleted. what you do with yours is your call

[1nvest]

Just been looking at the figures for using the bucket choice, 37% into 50/50 stock/gold accumulation/growth, 62% into a Permanent Portfolio as a 'inflation bond' type choice, and drawing 4.8% SWR from that PP (which in being 62% weighted = 3% overall SWR).

Worked well for a range of US based data start dates I looked at since the 1970's, and i also compared it against Terry's accumulation HYP with the same 3% SWR applied since 2001 when HYP was also started, and it pretty much trashed both (2.5% higher annualised). Basically whilst the growth bucket grew at a good rate, the PP as a drawdown bucket has retained its original inflation adjusted start date value after 20 years of a 4.8% SWR rate of 'drawdown'.