Role of Cash in the PP

[sophie]

This is how I look at the VP.  I put a certain percentage of my income into the PP knowing that it's safe.  The rest left over after living expenses can go into my VP which I understand to mean my own career and business. 

+1!!!  And same for HB's rule #1.  That's exactly why I quit tinkering with a VP:  it makes far more sense for me to put the effort into my career than into amateur trading.  I will bet that's true for a lot of people - though of course not everyone.

Dualstow - YES!!!  Remember that thread about the role of cash in the PP?  Until I started playing with spreadsheets I didn't appreciate how having this uncomfortably large chunk of "unproductive" investments can make such a difference to maintaining a portfolio's value during retirement.  It just goes to show that HB knew what he was doing, and everything in the PP really is there for a reason.

[iwealth]

I still don't understand this concept that withdrawing from cash during retirement in some way increases the "safety" of the portfolio. I've also tinkered with spreadsheets and found that withdrawing equally from all 4 asset classes is just as safe and depending on the start date of the backtest, often leaves you with higher portfolio values. It really all depends on the start date of the portfolio - you can't only backtest from one single day in 1972 and claim the results of that backtest provide all that much predictive value.

If anything, if we see some extended craziness from fed tapering and stocks/bonds/gold drop concurrently (as they did for awhile in 2013), withdrawing from cash only would be a terrible idea.

That said, I'd love to be corrected on this.

[Pointedstick]

I still don't understand this concept that withdrawing from cash during retirement in some way increases the "safety" of the portfolio. I've also tinkered with spreadsheets and found that withdrawing equally from all 4 asset classes is just as safe and depending on the start date of the backtest, often leaves you with higher portfolio values. It really all depends on the start date of the portfolio - you can't only backtest from one single day in 1972 and claim the results of that backtest provide all that much predictive value.

Fascinating. I just did this myself, starting from 1972 and I also discovered is that it's sort of a wash, but with the variable being the withdrawal rate. 2 or 3% withdrawal rates worked out better for cash, while higher rates (especially crazy high rates) made the cash withdrawal more risky:

[img width=500]https://i.imgur.com/4ffcZUT.png[/img]

Another thing was that withdrawing from cash triggered 13 rebalances, while withdrawing from the four equally only triggered 10.

Darn it, now I'm rethinking the importance of cash. In is the successful salaried retail investor a myth?, we concluded that cash was important for buffering down years. But here it looks like the problem with selling falling assets is helped by also selling rising or flat assets at the same time, so they generally even each other out… which also makes sense given the PP's risk-parity-style approach.  As long as one volatile asset's losses are usually being offset or broken even by gains in the other two, selling slices of all three at the same time isn't going to destroy the performance, and rebalancing ensures that you maintain a roughly 25x4 allocation.

I wonder if cash vs non-cash withdrawal being mostly a wash is a quirk of the historical time period favoring the PP, and if withdrawing from cash might be safer in a less favorable environment (e.g. gold and bonds in the toilet for 10 years, stocks don't keep up)

Now I kind of want to see a 1972-2014 backtest of 25x4 doing withdrawals from cash, all four, or the highest performing asset that year, and then a 1972-2014 backtest of 33x3, doing withdrawals from all three or the highest performing asset.

I'm pooped right now though. Maybe later.

[Pointedstick]

Okay, a cashless PP definitely does not work when you keep the same withdrawal rate and withdraw from all assets equally. Boning up on my Excel-Fu a bit…

[iwealth]

Now I kind of want to see a 1972-2014 backtest of 25x4 doing withdrawals from cash, all four, or the highest performing asset that year, and then a 1972-2014 backtest of 33x3, doing withdrawals from all three or the highest performing asset.

Would also be very interesting to see the results of those same backtests started other years. Even starting in 1983 gives us a full 30 year backtest. Starting the backtest in other years is just as important as testing withdrawals from the four different assets.

Okay, a cashless PP definitely does not work when you keep the same withdrawal rate and withdraw from all assets equally. Boning up on my Excel-Fu a bit…

A cashless PP probably has much higher max drawdowns so that doesn't surprise me at all. Doubt we'd see the same thing if instead of removing cash, you replaced the 25/25 bond/cash portion with 50% intermediate bonds.

[Tyler]

You guys inspired me to run a few numbers.  This is how the 10-year rolling maximum portfolio-preserving withdrawal rates look for various drawdown methods.  Both the Cash Drawdown and Total Drawdown methods used the 35/15 bands, but the Total DD withdrew money from each asset evenly.  The No Cash method rebalanced every year between only Stocks, Bonds, and Gold. 



The thing you want to look for on this chart is the low points.  The idea is that the max withdrawal rate that maintains your inflation-adjusted portfolio value over the worst 10-year historical period should theoretically last forever.  For that reason, I wouldn't recommend the No Cash option.  Historically, the added volatility offers more downside than upside for a retiree. 

I was surprised putting this together that the Cash DD and Total DD methods were so similar.  The low years have similar SWRs, and in many years the Total DD is marginally higher than the Cash DD.  Then again, any small benefit is likely negated by the tax advantages of using the Cash DD method. 

TL;DR : cash in the PP raises the portfolio-preserving withdrawal rate by almost a full percent, but over-thinking the drawdown method beyond being tax efficient isn't going to buy you any more safety. 

[iwealth]

That's great stuff Tyler, thanks. Any chance you can quickly modify your work to analyze the available 20 or even 30 year drawdown periods?

How/when does the cash DD method eventually go about rebalancing into cash? End of year, assuming it is below 15%? I'm not convinced of much of a tax benefit since as PS mentioned, the nature of risk parity asset classes tend to leave you with some capital losses each year to balance out the gains. And eventually you need to sell assets to rebalance into cash anyway so there'll be plenty of taxable events using that strategy as well.

[Tyler]

That's great stuff Tyler, thanks. Any chance you can quickly modify your work to analyze the available 20 or even 30 year drawdown periods?

How/when does the cash DD method eventually go about rebalancing into cash? End of year, assuming it is below 15%? I'm not convinced of much of a tax benefit since as PS mentioned, the nature of risk parity asset classes tend to leave you with some capital losses each year to balance out the gains. And eventually you need to sell assets to rebalance into cash anyway so there'll be plenty of taxable events using that strategy as well.

Perhaps I'll look at longer timeframes at some point.  But I should point out that the way I calculated it is actually more conservative than traditional 30-year SWR calculations.  Success for them is not running out of money in 30 years, even when there's a really bad stretch in the middle.  Success for me is not losing inflation-adjusted principal even during the worst decade, which means in other periods your portfolio is always growing.  IMHO, consistent real returns will keep you from abandoning the plan at the worst possible time. 

My spreadsheet used the 15/35 rebalancing bands and checked each asset once a year.  Interestingly, if someone retired in 1972 and used the bands with the Cash DD method, they had 15 rebalancing events through 2012.  Only 8 of the rebalances were triggered by cash falling below 15%.

The nice thing about using the Cash DD method is that in the 41 years I have numbers on, you only have to sell stocks, bonds, or gold in 15 of them.  Yes, the rebalancing events will be larger but you'll have several years to plan how to handle it.  One could also use the Total DD method in years when it is advantageous for any particular reason based on your own personal situation with no argument from me.

[Tyler]

Supporting my point about PP vs Boglehead portfolios when you discuss SWRs:



This really shows how the stock bull market in the 80s and 90s spoiled people.  But imagine retiring in 1971 with a recommended 4% WR, and watching your portfolio lose 6.5% a year for a full decade.  Those are the times that will make people forget the boom years, and few will wait around for some 30-year rolling average to save them based on a dusty old retirement study.  Cash and Gold help a lot.

[barrett]

Thanks for posting those charts Tyler. I have printed them out so that I can review them periodically from time to time when I am 100 years old and trying to remember what a SWR on the PP is.

[Tyler]

Thanks, Tyler! One question: Why does your chart end in 2003?

The chart shows 10-year rolling WRs based on the starting year.  So the 2003 data point covers the investment period from 2003-2013.  Anything after that would have incomplete data so I left it out.

[sophie]

Wow - quite an important hole in the argument!  We definitely should have tested with/without cash and using different withdrawal methods.

I fired up the spreadsheet (using CraigR's performance #s & CPI data from BLS.gov) and tested the 33x3 version, starting from 1972.  There were 7 rebalances, and with a starting $1M portfolio and 4% withdrawal (increased each year by the inflation rate), the ending balance was $14,338,467.  I assumed that withdrawals came equally from the 3 assets each year.  This is less than the >$16,000,000 value of the 25x4 portfolio (yay cash - but I should probably check that figure because it looks a bit improbable). It significantly outshone the 50/50 BH portfolio ($2M) or the 100% stock portfolio ($4M). 

As Tyler's chart makes clear, this has a lot to do with the start date.  1972 gave the PP a big advantage because of the huge jump in gold in the first 2 years.  Also note that the PP (either version) would be expected to do worse than the BH 50/50 if we skipped to 1981 as the start date.  Maybe later tonight I'll try varying start dates - but either way, this is backtesting and there are no guarantees.  The era we are in of very low interest rates is something that is not going to be reflected in the backtests, for example - so while this is fun and all, I would be careful about making decisions based on them.

[sophie]

OK, I checked a 1982 start date, which I imagined would be the nightmare scenario for the PP.  1982 started its worst performing decade with two losing years in a row right off the bat, and 4 losing years in the first 8.  Also found a bug in my spreadsheet - clearly my spreadsheet-Fu is not up to Tyler's level - so yes, 33x3 PP outperformed the 25x4 for both 1972 and 1982 start dates.  Forgot to mention above, I used 20/45 rebalance bands for the 33x3 PP (equivalent to 15/35 for the 25x4).

(drum roll) - no matter what, using a starting 4% withdrawal rate adjusted annually for inflation, either the 25x4 or 33x3 portfolios did great.  The 25x4 portfolio ended with a balance of over $4.2M, and the 33x3 with a balance of over $5.6M. The low point for the 33x3 was $950K, and for 25x4 it was $989K.

So I would say:  the classic PP is not only about as safe as it gets, even if you are underperforming your Boglehead friends, but the cash allocation is worth it for the added peace of mind factor.  I imagine it would also greatly simplify the withdrawal process.  And, you will still come out ahead in the long run.  What's clear from these backtests is that someone simply following the PP and not trying to tinker with it to improve returns is going to beat the pants off of 99% of investors.  Who ends up with a portfolio 4 x the starting value after 30 years of retirement????

[Roberto]

Thanks for the great work, Sophie.  It is really helpful to me as a PP newbie to see that folks are checking this out from every angle, and that the generally good results prevail regardless of the time frame.
You mentioned the fact that we are in a period of low interest rates, and it appears to me that we may be for some time to come.  I am wondering if anyone has looked at the possibility of eliminating bonds instead of cash for the 33x3.  With the bond bull market behind us, and the economy resting in the hands of the printophile Fed, I personally have grave doubts that the 3% return on 30 year treasuries is worth the risk of holding them.  Cash, on the other hand, keeps the door open to all kinds of possibilities.   

[goodasgold]

I am wondering if anyone has looked at the possibility of eliminating bonds instead of cash for the 33x3....  With the bond bull market behind us, and the economy resting in the hands of the printophile Fed, I personally have grave doubts that the 3% return on 30 year treasuries is worth the risk of holding them.

For years now, many people have been predicting a rise in interest rates. Yet year after year the rates have remained low. It stands to reason that rates have to go up eventually, perhaps very soon, but only Mr. Market knows for sure, and as usual, he ain't telling nobody. But the market for crystal balls is as hot as ever.

There is no escaping the old saying: "The market can remain irrational a lot longer than we can remain solvent."

And there is always the chance that rates will actually fall, making long bonds a winner. 

[dualstow]

Roberto, this thread was already carved out of another because it was off topic, but the 2nd thread from the top in the Bonds section is Why Hold Long Bonds Now.

[iwealth]

The nice thing about using the Cash DD method is that in the 41 years I have numbers on, you only have to sell stocks, bonds, or gold in 15 of them.  Yes, the rebalancing events will be larger but you'll have several years to plan how to handle it.  One could also use the Total DD method in years when it is advantageous for any particular reason based on your own personal situation with no argument from me.

The few rebalance events using the Cash DD method was partly contingent on cash's decent rate of return. Anyone starting their drawdown today would probably find themselves rebalancing into cash rather often, or at least I'd hope so...otherwise I'd hate to see the rest of the portfolio.

That's more or less my concern about the PP as a whole, at least in today's environment.

[barrett]

Agreed on that, iwealth, but we'll all be looking at cash differently if interest rates go up considerably. It's just so hard to know what the next "economic cycle" will be. We are all obviously used to (tired of?) no returns on cash but that is not the historic norm. I am hoping to start withdrawing from my PP/retirement accounts in about ten years and am hoping that cash yields something by then.

If cash yields do stay really low, and the PP is still delivering positive real returns, then that means the other three assets are doing well.

[sophie]

The few rebalance events using the Cash DD method was partly contingent on cash's decent rate of return. Anyone starting their drawdown today would probably find themselves rebalancing into cash rather often, or at least I'd hope so...otherwise I'd hate to see the rest of the portfolio.

That's more or less my concern about the PP as a whole, at least in today's environment.

Actually, the rate of return on cash relative to inflation is what's important.  Most of the time, it doesn't beat inflation (see 1970s).  Either way though, rebalancing into cash after multiple withdrawals is a problem why?  You'd certainly have the same issue with a 50/50 Boglehead portfolio (which owns no cash).

I like to think of the PP as a pigpen with 4 feeding troughs.  If one trough doesn't look good, the pigs run to one of the others.  Cash looks awful right now because of the low interest rates, and I bet that is what's keeping stocks going on such a tear:  dividend rates are higher than CD rates right now so it just makes more sense to buy dividend stocks or an S&P 500 index than CDs.  If you know exactly when that trend will reverse, then you don't need to hold cash.  I'm really bad at predicting such things, so I hold cash.

[Mark Leavy]

I like to think of the PP as a pigpen with 4 feeding troughs.  If one trough doesn't look good, the pigs run to one of the others.  Cash looks awful right now because of the low interest rates, and I bet that is what's keeping stocks going on such a tear:  dividend rates are higher than CD rates right now so it just makes more sense to buy dividend stocks or an S&P 500 index than CDs.  If you know exactly when that trend will reverse, then you don't need to hold cash.  I'm really bad at predicting such things, so I hold cash.

Just a quick note to thank you for the consistently high quality and thoughtful posts that you make to this forum.
You and a few other commenters make me feel like I'm surrounded by expert musicians.  I recognize the brilliance, but humbly appreciate that I will never reach that level of competence.

[murphy_p_t]

I still don't understand this concept that withdrawing from cash during retirement in some way increases the "safety" of the portfolio. I've also tinkered with spreadsheets and found that withdrawing equally from all 4 asset classes is just as safe and depending on the start date of the backtest, often leaves you with higher portfolio values. It really all depends on the start date of the portfolio - you can't only backtest from one single day in 1972 and claim the results of that backtest provide all that much predictive value.

Fascinating. I just did this myself, starting from 1972 and I also discovered is that it's sort of a wash, but with the variable being the withdrawal rate. 2 or 3% withdrawal rates worked out better for cash, while higher rates (especially crazy high rates) made the cash withdrawal more risky:

[img width=500]https://i.imgur.com/4ffcZUT.png[/img]

Another thing was that withdrawing from cash triggered 13 rebalances, while withdrawing from the four equally only triggered 10.

PS...I'm curious why the almost remaining balance at 3.5% vs. 4% being a substantial balance ?  any thoughts?

[murphy_p_t]

Tyler...thanks for your work here. My question is about the withdrawal rate in your charts...is that rate the annual fixed percentage of the portfolio? Or is it a starting rate adjusted for inflation?

[Tyler]

Tyler...thanks for your work here. My question is about the withdrawal rate in your charts...is that rate the annual fixed percentage of the portfolio? Or is it a starting rate adjusted for inflation?

It's a starting rate adjusted for inflation each year using actual CPI-U numbers to maintain purchasing power over time.

[Tyler]

[img width=500]https://i.imgur.com/4ffcZUT.png[/img]

PS...I'm curious why the almost remaining balance at 3.5% vs. 4% being a substantial balance ?  any thoughts?

The other interesting thing to me is that the ending values for the higher withdrawal rates dance around zero and don't get progressively more negative.  I can see some variability based on the timing of rebalances, but would figure a 6% WR would look a lot worse than shown.  In any case, Pointedstick's table is a good lesson is to err on the conservative end where you're less susceptible to statistical chance and rounding error. 

[Pointedstick]

I think the reason for the wonky table is because different withdrawal rates triggered rebalances during different times, and so inevitably some of these would have been at better or worse times relative to the others.