Worst 3 year PP performance ever?

[jason]

If anyone is not familiar with this, there was an article in Time Magazine about it in 2013, which I highly recommend:
http://business.time.com/2013/03/12/if- ... p-so-much/

That article seems misleading to me. Gold rising in price is no more inflation than stocks or bonds rising in price would be. Gold is an investment product, not a consumer product. Same with the commodities he describes. What do I care that the price of cocoa has gone up if the price of chocolate hasn't? The point about gas price inflation seems archaic today, as it's now fallen to below the level indicated in the article. And I don't understand how he can claim that taxes have gone up 9% a year for four years. Did I miss something? I certainly haven't seen my tax rates go up 35% in the past 4 years.

His point about wage stagnation reducing potential purchasing power and keeping price inflation low seems like a correct one, but that's a point orthogonal to inflation. It seems to contradict his point: if wage stagnation is keeping inflation low, how can he argue that inflation is actually higher than reported?

The nature of averages ensures that some costs are going up faster than the rate of inflation, and some are going up slower, or even falling. It may well be that inkjet printer ink (a real racket; buy a laser printer!) is rising faster than CPI, but simultaneously, gas prices may be plummeting.

I don't think the law of averages necessarily applies to the CPI because the formula may have been changed in a way that intentionally depresses the number.  I have not verified this, but I have read that the formula changes included under-weighting and/or removing gas and food from the equation, and overweighting clothing and electronics.  Gas and food are a substantial portion of many family budgets.  Electronics and clothing are probably not nearly as big of a part of a typical family budget. 

We know by walking around a Wal-Mart that food prices are far higher than they were 10 years ago, while electronics and clothing prices are around the same as they were 10 years ago.  I just bought a big screen TV that was bigger than the one I bought 10 years ago, and the new one cost less than the old one (in nominal terms, not real).  And you can still buy a t-shirt at Wal-Mart for $7.  But this isn't due to a lack of inflation.  This is due to extremely low overseas labor and manufacturing costs.  So, if certain types of goods made overseas are being cherry-picked and are overweight in the CPI, the number becomes skewed, and has less meaning.

I look at the CPI like the unemployment rate.  The formula for the unemployment rate has also been changed over the years, and anyone who is unemployed for more than 18 months is removed from the statistic.  These official statistics have less meaning when they are not reflective of what is happening in the real world.

Gold is actually a somewhat reasonable measure of inflation over time because gold has maintained its approximate buying power over thousands of years.  A well know example is that thousands of years ago, an ounce of gold could buy you a nice suit, a nice belt, and a nice pair of shoes.  Today, an ounce of gold can get you the same thing.  Of course, the price of gold fluctuates, so the the outfit you would have bought in September of 2011 for an ounce of gold would be quite a bit nicer than an outfit purchased today.  But, for example, if gold were trading at $10,000 per ounce, today, or $100 per ounce today, then it would be very apparent that neither of those prices are anywhere close to the value of a nice suit, belt and shoes.  Obviously, "nice" is a subjective term, but I would assume it means at the higher end of the market, and not made in China.

[MachineGhost]

As you wish. 

Why are you using a 2-year T-Note and not a 1-year or less T-Bill?  That bugs me.  The duration is high enough to make a different in tight money scenarios.

[sophie]

Tyler, this is absolutely gorgeous work.  I haven't had time recently (insert appropriate Princess Bride quotes here) so I'm really glad you did, because this is exactly what we all need to remind ourselves of after recent discussions.

Can I ask, how are you calculating the comparison to real returns?  Are you computing annual CAGR and then subtracting the CPI?  I wonder if that is actually not taking "variance drag" into account, which would favor the PP even more.

If your investments are 20 to 25 times your living expenses (4%SWR) then having 25% of them in cash gives you 4 or 5 years of cash.  What a system.

YES.  I was thinking exactly the same thing.  Further note, I've seen lots of articles recommending 5 years of living expenses to be held in cash in retirement.  Thus, if someone has a standard Boglehead portfolio at 25x expenses in retirement, their cash allocation is functionally similar to PP's.  So great to have it baked right into the plan.

[Tyler]

Why are you using a 2-year T-Note and not a 1-year or less T-Bill?  That bugs me.  The duration is high enough to make a different in tight money scenarios.

Just because it reflects my personal PP and the data Craig uses here:  https://web.archive.org/web/20160324133 ... l-returns/

Here's one for the T-Bill purists.

[img width=600]http://i60.tinypic.com/2na8mdv.jpg[/img]

It does knock returns down slightly (as you'd expect).  For reference, the worst 3-year CAGR is -0.3% for a total return of about -1%.That outlier 4-year negative is -.01% (basically zero). 

[Kevin K.]

As beautiful as it is edifying Tyler - thank you!

The next step I'm trying to do on my own - without a fraction of your Excel skills! - is comparing these results to other kinds of "low fat tail" portfolios that I think are more useful alternative choices for the kind of defensively-minded investor who'd be drawn to an allocation like the PP. Desert's 60:30:10 (10 Year Treasury:S&P 500:Gold) fares very well in such a comparison, for example, while avoiding the big allocations to gold and long Treasuries that keep many from investing in - or staying with - the PP. I can't imagine anyone looking at the PP even considering an allocation that has more than 50% in equities.

[Tyler]

Can I ask, how are you calculating the comparison to real returns?  Are you computing annual CAGR and then subtracting the CPI?  I wonder if that is actually not taking "variance drag" into account, which would favor the PP even more.

Good question.  I mentioned earlier I changed the inflation calculation.  My old method subtracted CAGR CPI-U from CAGR investments, which always bothered me precisely because of the potential variance drag. The new one as of the last few charts calculates the real return every year along the way before doing anything else, so it should be a lot more accurate. 

[barrett]

The next step I'm trying to do on my own - without a fraction of your Excel skills! - is comparing these results to other kinds of "low fat tail" portfolios that I think are more useful alternative choices for the kind of defensively-minded investor who'd be drawn to an allocation like the PP. Desert's 60:30:10 (10 Year Treasury:S&P 500:Gold) fares very well in such a comparison, for example, while avoiding the big allocations to gold and long Treasuries that keep many from investing in - or staying with - the PP. I can't imagine anyone looking at the PP even considering an allocation that has more than 50% in equities.

I really like the "Desert Portfolio." I do wonder about it not having a cash component though one could argue that with 60% ten-year notes it should spit out a lot of income each year. But that makes it potentially less tax efficient too. Might have to start another thread.

[Tyler]

The next step I'm trying to do on my own - without a fraction of your Excel skills! - is comparing these results to other kinds of "low fat tail" portfolios that I think are more useful alternative choices for the kind of defensively-minded investor who'd be drawn to an allocation like the PP.

[img width=350]http://i57.tinypic.com/6tg2vr.jpg[/img]

My dirty little secret is that since the start of this thread I figured out how to fully automate the process. As long as the returns data is available, this is easy. 

If anyone has a special request, send me a PM.  No promises on a fast response but I'll see what I can do.  If I get enough requests, I may start a new thread to share results.  Eventually I'm looking into building a website to consolidate this kinda info and satisfy my data fetish, so feel free to also PM me what data you find most interesting or would like to see visualized.

[Kike Moreno]

Really nice graphs Tyler, thanks a lot!

Since you ask for ideas/requests, we would like to see similar plots for the Euro PP described here:  http://www.carterapermanente.es/evoluci ... ermanente/

Its composition is:
25% MSCI EMU
25% Germany Gov 30y
25% gold
25% Germany Gov 1y

The timeframe will only be 16 years but it should be enough to see how it looks...

[ochotona]

I really like the "Desert Portfolio." I do wonder about it not having a cash component though one could argue that with 60% ten-year notes it should spit out a lot of income each year. But that makes it potentially less tax efficient too. Might have to start another thread.

Try backtesting Desert vs. a PP/Desert Hybrid... instead of 60% 10-year Treas, 30% ST Treas and 30% LT Treas. I think you'll like that combination, too.

[Kevin K.]

Tyler's cool tool makes backtesting like this great fun, but of course there are any number of bond-heavy allocations one could run that produce similar numbers. Larry Swedroe alone has at least a half-dozen iterations of his "low fat tails" portfolios, the most recent of which use 70% IT Treasuries and 30% small cap value international and EM equities to deliver 60:40 CAGRs with PP-like stability during market crises.

At the risk of hijacking the thread even further, isn't one aspect of the brilliance of the PP its formulation to respond to macroeconomic conditions rather than being purely a product of backtesting? There's no arguing with the historical robustness of the PP, but it seems to me (heretical as it may be to some) that there's always some room to ask if the 4 assets in the PP are still appropriate or need tweaking in light of changes in the market and in the world since Browne's time. HB himself of course altered the composition of the PP over time, as have Craig R. and others (e.g. substituting TSM for the S & P and short-term treasuries for the Treasury MM account, though these are admittedly minor tweaks compared to the difference between PPRFX and the 4 x 25 PP).

Three areas of major change come to mind: the unforeseen rise of paper trading in gold and unprecedented levels of behind-the-scenes manipulation of the gold price; growth of non-U.S. equities to ~55% of the total investable assets; ongoing willingess of our elected representatives to cause U.S. Treasury bond ratings to be tarnished and to cause the world to question the reliabiilty of "full faith and credit."

Obviously there's room for an infinite number of responses to such changes, ranging from the very valid choice of doing nothing because the 4 x 25 strategy has worked so well in the past (which is of course justifying future allocations through backtesting just as the Bogleheads do), to reducing or eliminating the gold allocation, increasing the equities slightly [from the gold] while diversifying them radically (Int'l, SCV and EM tilts) or conservatively (substituting VT for VTI) and making similar changes in the bond allocation (ranging from going with all IT Treasuries to switching to, say, 50% IT Treasuries, 25% TIPs and 25% hedged foreign bonds).

Some would mistake these kind of changes for market timing when they're really tactical asset allocation in light of macroeconomic conditions - which is of course exactly what the PP allocations Browne came up with originally were based on - in his time.

Rick Ferri has an interesting recent blog post on this relative to international equities:

http://www.rickferri.com/blog/investmen ... ri+Blog%29

[MachineGhost]

Rick Ferri has an interesting recent blog post on this relative to international equities:

http://www.rickferri.com/blog/investmen ... ri+Blog%29

What his long-term track record for this "market timing"?

[Tyler]

Really nice graphs Tyler, thanks a lot!

Since you ask for ideas/requests, we would like to see similar plots for the Euro PP described here:  http://www.carterapermanente.es/evoluci ... ermanente/

Its composition is:
25% MSCI EMU
25% Germany Gov 30y
25% gold
25% Germany Gov 1y

The timeframe will only be 16 years but it should be enough to see how it looks...

Hmm... International portfolios are currently out of my wheelhouse. But I'll keep the request in mind if I expand data sets.

[MachineGhost]

We got a new one from Tyler:



The lack of T-Bonds really hurt Swedroe in the 1973-1974 bear.  Or was it gold?

[Kevin K.]

Nice to see that comparison - thanks MachineGhost and Tyler!

As for what hurt the Swedroe portfolio, I'd say it's more a matter of the once-in-a-lifetime events that helped the PP. Quoting William Bernstein:

"For starters, gold was not easily investible during the first decade of this period; in fact, it was downright illegal to own the stuff from 1933 to 1974. Start the analysis in 1980, for example, and you were a full percent better off leaving it out of the portfolio entirely (i.e. owning one third stocks, bonds and bills). Put another way, you can go off the gold standard and open up its ownership only once. It seem highly unlikely that gold will return several percent more than inflation in the coming decades; almost by definition, zero percent above inflation seems more like it."

To be clear, these are Mr. Bernstein's views, not mine. One could easily make a parallel point about SCV and Emerging markets not really being investible assets until very recently.

[mukramesh]

@Kevin: That's true, but gold also seemed to help the PP in the 2000's. I mean, look at all that green compared to the Swedroe portfolio! Also protection against 2008-like events is pretty nifty

[Tyler]

I really like the "Desert Portfolio." I do wonder about it not having a cash component though one could argue that with 60% ten-year notes it should spit out a lot of income each year. But that makes it potentially less tax efficient too. Might have to start another thread.

Try backtesting Desert vs. a PP/Desert Hybrid... instead of 60% 10-year Treas, 30% ST Treas and 30% LT Treas. I think you'll like that combination, too.

[MachineGhost]

I think Desert & Swedroe data-mined the 80's and 90's to come up with their portfolios.

[MediumTex]

Even after gold's run up in the 1970s, gold provided key support to the portfolio during the 1982-2000 bear market for gold in years when the other assets performed poorly.  Look at the year by year PP results for the 1982-2000 period and you may be surprised to see that gold was the leader in the portfolio in a few of those years.

An asset in a secular bear market can still be traded for gains if you can buy low and sell high, which is what the PP forces you to do when you rebalance.

[Tyler]

You really need to delete the first few years when it was illegal to own gold in the U.S.

One can gray out the first few rows of the PP chart if they like (they were pretty average, anyway, and wouldn't change overall impressions), but the harsh times for other portfolios in the 70's are still real. 

I do personally like seeing how gold would have affected various portfolios during that time even if it was illegal for a while.  It's not anymore, and is a nice tool for addressing the economic conditions in the early 70s should they ever arise again. 

[MachineGhost]

You really need to delete the first few years when it was illegal to own gold in the U.S.

Why?  We want to see the performance under as much history and environments as possible.  And gold was able to be legally purchased starting in the early 60's through derivatives or holding offshore, besides already legal gold jewelry and [semi]numismatic gold coins (TR was a coin collector, so FDR made a crony exception).  Bullion coins weren't domestically legal again until 1975.  At the minimum, silver was the poor man's gold.

I mean, did you think that gold being illegal stopped HB and all the other gold bugs?  He bought before 1970.

[lordmetroid]

How about if you was a european and had to deal with the additional risk of the dollar value?
Personally as a Swede I am very interested in seeing these assets would perform in SEK:

Bank account( SEK )
Swedish OMX30( SEK )
TLT( USD )
Gold( USD )

[MachineGhost]

The 30/60/10 fits me a bit better because I think the real return will be a bit higher, and 10% physical gold holding is practical. Being the tinkerer that I am, I've gone a bit further and have introduced a higher risk/higher expected return mix into the equity side, similar in concept to a Larry portfolio.  It backtests well, but may or may not perform well in the future. I wouldn't recommend it to anyone but myself. 

The #1 problem with your portfolio being 60% in T-Bonds is it will underperform the PP in a bond bear market or high inflation regimes, which is not fully represented in the data since the currencies were floated.  We can glimpse a small taste of what to expect in terms of real returns during the tail end of the last Great Inflation:

Code: Select all

Year	PP	Desert
1968	5.36%	2.12%
1969	-9.59%	-12.31%
1970	-0.33%	3.37%
1971	7.60%	8.16%
1972	14.62%	10.41%
1973	5.85%	-10.11%
1974	-0.11%	-12.80%
1975	-0.79%	4.57%
1976	5.96%	12.59%
1977	-1.39%	-5.54%
1978	2.81%	-4.30%
1979	24.63%	3.85%
1980	0.78%	-4.46%
1981	-14.10%	-13.59%

[MachineGhost]

MG, take another look at your data ... seems to be a problem, at least in 1980 and 1981. 

But yeah, if we repeat the 1970's, where we went off the gold standard in 1971, then legalized gold in '75, yes, the PP is preferable in those conditions.  100% gold is even better.

Checked, data is good.  This is what I have for nominal returns:

Code: Select all

Year	Stocks	T-Bills	T-Bonds	Gold
1980	31.74%	11.25%	-4.98%	15.19%
1981	-4.70%	16.58%	0.01%	-32.60%

[MachineGhost]

That's pretty close to what simba's spreadsheet has as well.  What do you have for 10 year bonds?  Or were you using a 50/50 T-Bill/T-Bonds for the 30/60/10 portfolio?  The 50/50 barbell results in a term and duration considerably longer than a 10YT.  I'm modeling a 10YT for the 30/60/10.  It helps a bit in rising-yield environments, relative to a barbell with a longer term and duration.  I nearly went with 5YT, but the 10YT tends to balance the volatility of the equities better.

Oh, you should have been more clear about that as I know some people are following your model but using 30yr bonds?  Eeek!  I modeled mine and yours with 30yrs and 20yrs when 30yr wasn't available.

I think 10yrs @ 60% would certainly change things vs 30yr.  I don't have 10yrs but I think your gains and losses would be all correspondingly lower?

EDIT: Here is real with 10yr:

Year PP Desert
1968 5.36% 2.12%
1969 -9.59% -12.31%
1970 -0.33% 3.37%
1971 7.60% 8.16%
1972 14.62% 10.41%
1973 5.85% -10.11%
1974 -0.11% -12.80%
1975 -0.79% 4.57%
1976 5.96% 12.59%
1977 -1.39% -5.54%
1978 2.81% -4.30%
1979 24.63% 3.85%
1980 0.78% -4.46%
1981 -14.10% -13.59%