PP withdrawals in retirement

[Gumby]

I understand the desire to analyze these things and I don't think it's good to use the word "bonkers" to describe it. I'm just not sure backtesting is going to work here. It is more productive for me to look at economic cycles in various places and see how various assets performed when put under stress and good conditions respectively. It's not perfect, but you can come away with some interesting observations on portfolio construction even if the data is incomplete.

Perhaps "bonkers" was a bit strong. (Sorry!)

Still, the longer the duration of a flawed data set, the larger the error would be. In this case, having the entire world, and all its assets, denominated in a currency that was fixed to gold is not applicable to our current word where all of the assets (including the inflation-protection assets) are denominated in a free-floating currency. It affects everything. And the more years you draw out the flawed data set, the more irrelevant and diverging the conclusions become.

[Gumby]

On or off the gold standard IMO makes no difference. The baseline many investors measure relative to is not cash nor gold, but inflation. Whilst inflation is far from perfect, its the best broad/general measure of retention or loss of purchase power. The fact that gold might have stayed at a fixed price whilst stocks might have doubled in price over a period of time is largely meaningless without knowing what the combined value/weightings would actually buy you. If inflation had risen 50% then those nominal gains/losses are much more meaningful/informative. Relative to inflation the real value of gold both pre and post 1972 has been far from fixed.

Since cash was pegged to gold, then the price of any asset was affected by that peg — including all inflation protection assets. Inflation-protection assets wouldn't have exhibited the volatility that we would have needed for a PP. And the gold-pegged price of those assets affected everything else, stocks, bonds, homes, etc — which were also denominated in a gold-pegged currency. Chaos Theory shows us that even the tiniest error in your data would end up causing enormous errors the longer you draw out your results — rendering them mostly irrelevant.

Whether such observations are bonkers or not is a personal choice. Perhaps only bonkers if you prefer to narrow your view to a particular favourable curve fit to reinforce your beliefs.

We could say the same exact thing about your data and results — using flawed models and drawing errors out over many decades to prove a point that will never repeat itself in the same manner.

[rickb]

I know Clive is perfectly capable of defending himself, but I thought I might try to reframe the discussion here a bit.  Rather than take what he's saying as an attack on the PP, I hear him expressing a concern that there have been periods where the PP has not done so well.  Yes, the 1981-2005 time period is cherry picked but no more so than any example interval chosen to prove a point.  Does the PP do fine over most periods you can pick (from 1971 on)?  Absolutely - and I don't think Clive disagrees.  Are there periods where it hasn't done so well?  Well, clearly, yes.  I think the questions Clive is exploring are

1) what are the characteristics of periods where the PP hasn't performed very well

2) have similar periods occurred before (even before 1971)

3) is there anything that can be done to increase performance during such periods (that doesn't hurt during other periods as well)

These all sound like perfectly reasonable questions to me - questions that I suspect Harry Browne would be pondering if he were still alive.

Were the conditions at the end of 1980 unique in history (gold in a bubble, long term rates very high, stock market P/E very low)?  From there gold went down, long term bonds went up, and the stock market went up - but yet the PP didn't even match ST bonds for the next 15 years (i.e. during this time you would have been better off with 100% ST bonds rather than the PP).

How about year by year?  From 1972 through 2008, the PP underperformed ST bonds in

1980: stocks up, LT bonds down, gold up
1981: stocks down, LT bonds up, gold down
1983: stocks up, LT bonds up (but basically flat), gold down
1984: stocks up, LT bonds up, gold down
1988: stocks up, LT bonds up, gold down
1990: stocks down, LT bonds up, gold down
1992: stocks up, LT bonds up, gold down
1994: stocks down (but basically flat), LT bonds down, gold down
2000: stocks down, LT bonds up, gold down
2001: stocks down, LT bonds up, gold up
2002: stocks down, LT bonds up, gold up
2008: stocks down, LT bonds up, gold up

This is 12 out of the 37 years from 1972 to 2008 inclusive.  Is there anything these have in common?

It's once out of the 4 times stocks were up, LT bonds were down, and gold was up (80).

It's 3 out of 3 times stocks were down, LT bonds were up, and gold was down (81, 90, 00).

It's 4 out of the 11 times stocks were up, LT bonds were up, and gold was down (83, 84, 88, 92).

It's once out of one time all 3 assets were down (94).

It's 3 out of 3 times stocks were down, LT bonds were up, and gold was up (01, 02, 08).

Looking for commonalities, it's 7 out of the 10 times stocks were negative including all 6 times stocks were down and LT bonds were up (regardless of whether gold was up or down).

If the goal is "don't lose money", the standard PP does very well (two losing years since 1972).  If the goal is "beat ST bonds", the standard PP comes up short about 1/3 of the time - quite often if the stock market is negative for the year.  Can this be improved?

[Greg]

The time for caution IMO will be when real yields turn positive again.

Has anyone come up with a correlation of how a traditional PP does against real interest rates? If gold has a reasonable correlation to PP performance as Clive said, would a potential idea be to lessen the gold portion of the portfolio when we have positive interest rates and funnel that cash instead into ST bonds, LT bonds and stocks?

So perhaps for a negative real interest rate environment we have:
25/25/25/25 Stocks/LT Bonds/ST Bonds/Gold

For positive real interest rate environment we have (for example):
28/28/28/16 Stocks/LT Bonds/ST Bonds/Gold

Not sure if someone has looked into this yet.

[Gosso]

The time for caution IMO will be when real yields turn positive again.

Not sure if someone has looked into this yet.

Yep.

http://gyroscopicinvesting.com/forum/ht ... ic.php?t=3

I also have a theory that gold tracks global M2, except with massive volatility.  I don't have my excel charts with me, but based on global M2 growth, gold is priced exactly the same as it was back in 1968.  It has basically spent the last decade catching up to M2 after falling behind during the 80's and 90's.  I haven't looked at M1, but I'm guessing it's fairly similar.  I like this theory since it means that gold is still acting as an unofficial gold standard.  Also the chart shows that a spike up to $10,000 would be equivalent to the 1970's bubble, although it could go higher if M2 growth accelerates.