Is "bucketizing" a "MIRAGE" of safety

[rayvt]

it isn't so much bucketizing that works as it is the fact you are pulling spending cash early on from other places other than your stocks.

if you read the latest study by michael kitces and dr pfau jointly they found anything that lets your stocks grow longer without being liquidated increase your allocation to stocks as the stocks grow and the other money becomes less as it is spent down.

Having a significant "money bucket" is really just pre-selling stocks. It's not that you are "pulling spending cash from other places than your stocks", it's that you've pulled out (or never put in) money from stocks in order to have it in that other place.

One way you have $25K in cash bucket and $75K in stock bucket. The other way you have $100K in stock bucket. One way you have $75K growing, the other way you have $100K growing.

 

[Meadbh]

Having a significant "money bucket" is really just pre-selling stocks. It's not that you are "pulling spending cash from other places than your stocks", it's that you've pulled out (or never put in) money from stocks in order to have it in that other place.

One way you have $25K in cash bucket and $75K in stock bucket. The other way you have $100K in stock bucket. One way you have $75K growing, the other way you have $100K growing.

Yabbut....let's say you need to withdraw 4% ($4K) this year, the market has tanked by 50%, and your $100K is now $50K. You now have an 8% withdrawal rate and are left with $46K in equities and bonds. In order to recover before your next withdrawal, your fully invested portfolio must increase by $54K, or 117%. How's that workin' for ya?

Had you kept two years' worth of expenses ($8K) in cash, your original equity / bond portfolio would have been $92K. Post meltdown, it would be $46K, plus your $8K in cash, for a total of $54K. You can withdraw the $4K you need from the cash portion and allow the remainder to recover (hopefully) before tapping into it. You can even wait it out for a second year, while you make contingency plans.

The purpose of having some cash on hand is not to increase portfolio return, but to decrease the risk of portfolio ruin if disaster strikes. And that's why it helps people to sleep at night. Including me.

 

[timo2]

using the buckets to visualize how retirement funds are invested has been really helpful in aiding my DW to understand the various retirement investments and their purpose in the AA. That said, my withdrawal strategy is unique to our situation. These 'one size fits all retirees' shysters really just take advantage of people, IMO.

 

[samclem]

Most stock market downturns are relatively short-lived, especially if we include the gains from dividends. From a 2009 NYT article:

In fact, according to a Hulbert Financial Digest study of down markets since 1900, the average recovery time is just over two years, when factors like inflation and dividends are taken into account. The longest was the recovery from the December 1974 low; it took more than eight years for the market to return to its previous peak, which was reached in late 1972.

Some may prefer other approaches, but this is why a "cash bucket" has appeal for many people. Having about two years worth of cash would historically allow them to avoid selling equities at a loss during >most< downturns. More cash would allow them to wait out even slower recoveries--at the cost of overall reduced portfolio returns due to a lower allocation to equities or bonds.

By not rebalancing when the market is down (keeping the equity allocation higher as the prices revert to the mean), I can see how a moderate size cash bucket approach might even beat the performance of "standard" rebalancing (with the same starting percentage in cash). And before anyone says it--I'll deny that it is market timing!

 

[photoguy]

Surely some academic has tested a cash bucket approach in a reproducible manner?

 

[rayvt]

Right. And in the cases where the down market reaches it's third year, the cash bucket is empty and they have to sell stocks for *much* less than they'd have been able to get 1 or 2 years ago. Heck, what about a period like the 8 years after 1974?

Average of 2 years --- a 6 foot man can drown crossing a river that has an average depth of 3 feet.

I'm trying to picture somebody who is so afraid of a market drop that they purposely shift their asset allocation more and more toward stocks as the market continues to fall, ending up with 100% stocks when they empty the cash bucket. Not having much luck visualising such a person. I think that if they actually realized that this is what they're doing, then they wouldn't do it.


Anyway, all this is handwaving so far. Thing to do is model it in a spreadsheet. As it so happens, there's a spreadsheet here: https://www.dropbox.com/s/cwprtn6y8ouyajj/SPY_Withdraw_by_Guyton_rules.xls that has monthly data for S&P500. It shouldn't be too hard to use this as a base and model a cash bucket method.

 

[rayvt]

Surely some academic has tested a cash bucket approach in a reproducible manner?

I think somebody did, in a 2-page spreadsheet. Unfortunately he lost the spreadsheet and now the SEC just stomped him flat.

 

[Meadbh]

Surely some academic has tested a cash bucket approach in a reproducible manner?

Here are a couple of references from Google Scholar. The first one is a thesis and deals with aggregate Swiss pension funds, and concludes that the bucket strategy works. I don't have access to the second article, but the abstract makes the point that risk management should be a high priority.

http://e-collection.library.ethz.ch/eserv/eth:28197/eth-28197-01.pdf

An Error Occurred Setting Your User Cookie

 

[photoguy]

Right. And in the cases where the down market reaches it's third year, the cash bucket is empty and they have to sell stocks for *much* less than they'd have been able to get 1 or 2 years ago. Heck, what about a period like the 8 years after 1974?

I'd just combine cash and FI into one category. In which case, except for extreme portfolios or extreme withdrawal rates, one would could last for years or even more than a decade on cash + bonds.

I think somebody did, in a 2-page spreadsheet. Unfortunately he lost the spreadsheet and now the SEC just stomped him flat.

He had that coming. But I did qualify my request with "academic".

I'm actually not convinced that a "cash buffer" would be that different from a standard AA approach where you just pull money out of the asset class that is most above it's allocation. In MeadBH's example, cash would go from 8% to 16% of the portfolio so would be the first to be spent down anyway.

I guess the devil is in the details of how the cash bucket is refilled. It would be nice if there was a paper that explored various options (i'm too lazy / don't care enough to write one).

Here are a couple of references from Google Scholar.

Thanks for the links. The thing that worries me about the first link is the author states he only used eight years for the back-test (how did that ever pass his review committee?). I can't access the second either.

 

[mathjak107]

Right. And in the cases where the down market reaches it's third year, the cash bucket is empty and they have to sell stocks for *much* less than they'd have been able to get 1 or 2 years ago. Heck, what about a period like the 8 years after 1974?

Average of 2 years --- a 6 foot man can drown crossing a river that has an average depth of 3 feet.

I'm trying to picture somebody who is so afraid of a market drop that they purposely shift their asset allocation more and more toward stocks as the market continues to fall, ending up with 100% stocks when they empty the cash bucket. Not having much luck visualising such a person. I think that if they actually realized that this is what they're doing, then they wouldn't do it.


Anyway, all this is handwaving so far. Thing to do is model it in a spreadsheet. As it so happens, there's a spreadsheet here: https://www.dropbox.com/s/cwprtn6y8ouyajj/SPY_Withdraw_by_Guyton_rules.xls that has monthly data for S&P500. It shouldn't be too hard to use this as a base and model a cash bucket method.

in reality i don't think any of us would have the nerve to go 100% stock and no other spending money.

but if we did and sold stocks off in good or bad times the 100% equity allocation historically performed the best in just about every rolling 30 year time frame.

the weight of the other buckets held equities down so much in the good years that 100% equities even selling into occasional losses were stll the clear winner.

the higher your equity allocation even spending down into those losses the better the results.

so using buckets for the first 5 to 7 years and letting equities run lessens the chances of selling into losses early on and if stocks are up you let them grow without selling any.

i think the recent data from pfau and kitces show that buckets that allow stocks to grow early on work the best and even selling into losses still produces the better results as long as you had some good years mixed in.

if you keep the stocks the same and spend down other cash first your equity allocation rises by definition. higher equity allocations have always yielded better results.

 

[teejayevans]

Anyway, all this is handwaving so far. Thing to do is model it in a spreadsheet. As it so happens, there's a spreadsheet here: https://www.dropbox.com/s/cwprtn6y8ouyajj/SPY_Withdraw_by_Guyton_rules.xls that has monthly data for S&P500. It shouldn't be too hard to use this as a base and model a cash bucket method.

Did that, comparing all S&P vs S&P/Bond(cash) and withdrawing bond bucket first, the latter wins in every down market, the opposite is true if you start in an up market. I'm not worried about having too much money, so I focus on the down side. As someone once said, I'm more interested in the return of my money as oppose to the return on my money.
TJ

 

[WilliamG]

My first exposure to buckets like this was with Frank Armstrong articles on Morningstar in 2000, the year I retired. He proposed a non-equity allocation equal to 5-7 years of spending and that the non-equity be high quality short term bonds. Cash takes to the extreme of high quality short term and is probably more luring now due to concerns about bond market. I also think that Frank has extended this allocation to a recommended 7-9 years.

As someone mentioned above, the element not talked about much is the rebalancing you need to do as time goes along.

 

[teejayevans]

My first exposure to buckets like this was with Frank Armstrong articles on Morningstar in 2000, the year I retired. He proposed a non-equity allocation equal to 5-7 years of spending and that the non-equity be high quality short term bonds.

As someone mentioned above, the element not talked about much is the rebalancing you need to do as time goes along.

Right, his was a 2 bucket system, Ray added a 3rd:
1. Cash/ST bonds
2. Balance funds
3. Equities
That was his first book, then his second book he added a 4th bucket (REITs?)
and then the next book a 5 bucket (annuties), I'm pretty sure he was on his way to 7 before too long .
Let me give you a hint on rebalancing...now would be a good time
TJ

 

[rayvt]

Well, it turned out to be pretty easy to model this in a spreadsheet. Download link: https://www.dropbox.com/s/xf4ma5blug27aws/SPY_Withdraw_by_CashBucket_rules.xls
It compares withdrawals using a "cash bucket" and a plain standard 4% SWR.
You can set a few parameters: start year, SWR, # years of cash in the cash bucket, starting amount, and asset allocation weights.

Defaults: 1972 start (just before the 1973-4 bear), 4% SWR, 5 years in the cash bucket, allocation of 6 parts S&P500 and 1 part 10-yr Treasuries.

As is often the case in investing, a seemingly emotionally comfortable strategy is detrimental to your financial well-being in the long-term.

I welcome any critique of the spreadsheet, or comments about any errors in it.

 

[samclem]

I welcome any critique of the spreadsheet, or comments about any errors in it.

Thanks for the work, I'm retracing your steps so that I understand it. I couldn't find your assumptions for annual returns on the S&P--was it by share price alone or did it include dividends? And what was done with the dividends--into the cash bucket, held for the first thing to be spent the following year (since they've just been unavoidably taxed), or reinvested to buy more equities?

Graphing: This used to be so easy to do using the very nice wizards in Excel. The newer versions stink. Another Microsoft "upgrade".

 

[obgyn65]

I have never planned to use any bucket-based methodology. Sounds like a waste of time.

 

[Meadbh]

Thanks rayvt,you are the spreadsheet guru!

I see that it takes a few years before the portfolios diverge. That suggests that a long term bucket strategy is a bad idea, but it doesn't convince me to empty my bucket during the crucial first years of ER, when I see it as insurance against the black swan.

 

[Meadbh]

I have never planned to use any bucket-based methodology. Sounds like a waste of time.

Of course it's a waste of time if your cash bucket is 99%!

 

[rayvt]

I couldn't find your assumptions for annual returns on the S&P--was it by share price alone or did it include dividends?

That data came from another spreadsheet which very comprehensively computes the gains from the SPX monthly closing prices and the exact dividend yields by month. All dividends re-invested at the beginning of the next month. Note that I don't use Yahoo's "adjusted close" -- I compute the returns myself. Returns assume no commissions or taxes. Most brokers will DRIP for free. Taxes are an individual personal matter; inside an IRA there is of course no tax.
And in the comparison of CashBucket to standard SWR withdrawals it washes out -- the handling of dividends & costs is the same for both methods.

I see that it takes a few years before the portfolios diverge. That suggests that a long term bucket strategy is a bad idea, but it doesn't convince me to empty my bucket during the crucial first years of ER, when I see it as insurance against the black swan.

The acid test is a 1973 start date. You get hit right off the bat with S&P500 losses of -15% and -27% in the first 2 years.

It's not such much the slow divergence in the early years, it's the hidden rot that happening inside the portfolio because of the large allocation to cash.

Look what's happening for a 1973 start (4% SWR, 5 years of cash). The first 2 years you leave the port alone and draw down the cash. The the 3rd year has a large gain and you transfer a very large amount from the portfolio to refill the cash bucket. The catchup amount is those 2 years of draw plus the accumulated inflation. What you are doing is selling a lot of stock just when the market starts to move off the bottom, instead of selling a little bit each year on the way down.
[*]
Now going into the 4th year, the Buckets stock allocatio is $51K whereas the Standard stock allocation is $72K. The S&P return for the 4th year was 24%. 24% gain on $72K is a lot more money than 24% gain on $51K.

This happens over and over again. Each time the market dips you don't sell stocks -- thereby letting your losses run -- and draw from the cash bucket. Then when the market recovers you lickety-split pull money out of stocks -- thereby cutting off your winners -- to refill the cash bucket. This acts as a rachet pulling money out of stocks and moving it to cash. Which earns nothing.
[*] I just realized that what you are doing is holding on during the drop and then selling on the way up. This is what amateurs do. The correct thing to do is the opposite -- buy on the way up and sell on the way down.

 

[teejayevans]

I welcome any critique of the spreadsheet, or comments about any errors in it.

I had to put in 4 to get 5 years of cash...and starting in 73 (since its 1st down cycle) would make it a bit more favourable to bucket method, since its really to protect the downside, also, if you went back to 1929, the effect would be even more pronounced.
TJ

 

[teejayevans]

I just realized that what you are doing is holding on during the drop and then selling on the way up. This is what amateurs do. The correct thing to do is the opposite -- buy on the way up and sell on the way down.

The "correct thing" sounds like trying to time(chase) the market, I'll pass.

Also you don't sell on the way up....you are basically waiting for the market to return to normal, so you sell AFTER the market recovers, for example if you retired in 2008, you lived off the cash, it's now 2013, good time to rebalance.

One more thing, as your portfolio grows, the 5 years of cash as a percentage gets smaller...unless you tied your expenses to the size of your portfolio.
TJ

 

[Gumby]

This whole thread reminds me of the great debate between Calvinists and Arminians over whether grace is prevenient or irresistible.

That is all. Please carry on.

 

[yakers]

This whole thread reminds me of the great debate between Calvinists and Arminians over whether grace is prevenient or irresistible.

That is all. Please carry on.

I'll take grace anyway it shows up.

Same for good market returns

 

[WilliamG]

Here is an interesting article on buckets maintenance from Morningstar.

A Bucket Portfolio Stress Test

 

[rayvt]

Interesting article. Unfortunately a "stress-test" that is only 6 years long, with 5 up years and only 1 down year isn't much of a test, let alone a stress test.

Anyway, I've updated the spreadsheet. It now shows better graphs, and allows you to select different ways to invest the cash. Same URL: https://www.dropbox.com/s/xf4ma5blug27aws/SPY_Withdraw_by_CashBucket_rules.xls