Why "should" the PP system work?

[aaa]

Good morning, this is my first post in this forum  .

I have a question on the PP methodology.

Why should the PP on the long term give approx + 3% of real returns instead of, say, - 3% of real returns?

Do you know of some rationale other than the historical records? I think that trading or investing methodologies based on historical data usually loose their profitability as soon as enough people applies them. I think I remember this idea, o a similar one, is on the PP book.

Thank you for any ideas you could share.

[MediumTex]

Harry Browne first proposed a PP-like allocation over 30 years ago, and it has continued to work in the intervening decades, so I don't think that it's a strategy that only works based upon backtesting. 

As far as "why" the PP works, if you read our book and still don't have a sense for why the strategy works, I don't know what I could write in a single post that would help to answer that question for you.  Perhaps someone else will be able to provide you with some good feedback, though.

Welcome to the forum and thanks for posting your questions.

[Pointedstick]

1. Wide diversification across assets whose correlation changes over time
2. Rebalancing to automatically buy low and sell high
3. Ownership of assets that are more volatile rather than less to magnify the effect of rebalancing

[AdamA]

Why should the PP on the long term give approx + 3% of real returns instead of, say, - 3% of real returns?

Also, as HB notes in his book, winning assets tend to rise more than losing attends tend to fall. 

[aaa]

Thanks to all for such quick answers. I apologize for any grammatical or spelling mistakes, English is not my native language.

Medium Tex, by what you say "...our book...", I understand that you are one of the authors, Roland&Lawson, of the PP book. I consider a privilege to be able to talk about this topic with you.

I read the book, and the idea I extracted was that, being the four asset classes so different, the losses in some asset classes are almost always more than compensated by the gains in the others.

My doubt is in the word "more". It's clear that the variability of a portfolio consisting of uncorrelated (or even negatively correlated) assets is less that the individual volatilities, but I could see no reason why the overall expectation turns out to be positive consistently over time. AdamA, as you said, "winning assets tend to rise more than losing attends tend to fall", but apart from relying on historical data, I didn't see why that has to be so in the future.

I enjoyed very much the book, but it seems that I have not been able to grasp all of its ideas. Reasons 2 & 3 in Pointedstick's answer tend to answer my question, specially the buying low (what has fallen) and selling high (what has risen). But what if what has fallen keeps on falling and what has risen keeps on rising? What if what has falling doesn't go up as much as what has risen goes down? I see the general reason and the historical data, I just wanted to think if I was missing something else...

Many thanks again for any ideas

[Pointedstick]

I enjoyed very much the book, but it seems that I have not been able to grasp all of its ideas. Reasons 2 & 3 in Pointedstick's answer tend to answer my question, specially the buying low (what has fallen) and selling high (what has risen). But what if what has fallen keeps on falling and what has risen keeps on rising? What if what has falling doesn't go up as much as that has risen goes down?

It's a valid concern. But ask yourself this: if assets in your portfolio are falling more than they're rising, where is that excess money going? It's either going into cash, or leaving the country.

People going into cash is a short-term phenomenon because most investors prefer to take risks for greater reward, so that money is going to leave cash soon enough. And if the money is leaving the country and going into another country's stocks, you can to a certain extent mitigate that by holding an all-world stock index fund instead of a country-specific one. I don't do this yet but am planning to the next time I harvest some stock gains.

Now, iff an asset is going to zero, you have bigger problems. However, these events tend to be dramatic, so it's not like you're going to be continuously rebalancing into an asset that eventually dies. It's going to give up the ghost practically overnight, IMHO.

Bonds can go to zero, but this is going to be quick and sharp, and if this happens, your country is probably in ruins and what you need is real assets and real skills! Same for stocks. And unless space aliens arrive from a planet made of gold, gold is always going to have some value and I don't see it going to zero.

Hope that helps!

[rickb]

Each of the assets has a non-negative expected real return.  According to the numbers posted at http://www.bogleheads.org/wiki/Historic ... ed_returns the expected real returns are:

Cash (0-2 year Treasuries): 0.3%
Long Term (20 year) Treasuries: 1.5%
Total stock market: 5%
Gold: 0% (Bogleheads in general dislike gold, so don't list an expected return, but I think 0% is the general assumption for gold's expected return)

If we combine these by adding them and dividing by 4, we get only 1.7% - but that's not the right way to determine an expected outcome from a collection of assets.  A better way is to use Modern Portfolio Theory (MPT, see http://en.wikipedia.org/wiki/Modern_portfolio_theory) which uses the expected return, standard deviation, and correlation of individual assets to determine the expected return and deviation of a collection of assets in some proportion.  I don't know the exact real return predicted by MPT for the PP, by I'd guess it's in the 2% range.  This number includes the benefit of rebalancing.

So, where does the "extra" 1% come from?

Here's where Browne's genius comes into play.  MPT is based on some assumptions that don't actually turn out to be true - specifically that investment returns in a particular asset are random following a normal distribution curve and correlations between assets are fairly consistent over time.  In reality, investment returns are not predictable (but not random following a normal distribution curve) and correlations between assets can (and do) change dramatically.  The assets comprising the PP are chosen because they act in predictable (not random) ways in various economic conditions.  The "extra" 1% comes from this predictability.  Gold tends to go up in response to fears of inflation.  Long term bonds tend to go up during periods of deflation.  Stock tends to up in periods of prosperity.  I don't see any of these changing any time soon.

Medium Tex, by what you say "...our book...", I understand that you are one of the authors, Roland&Lawson, of the PP book. I consider a privilege to be able to talk about this topic with you.

MediumTex is J.M.Lawson.  Craigr (who also posts here) is Craig Rowland.

[aaa]

It's a valid concern. But ask yourself this: if assets in your portfolio are falling more than they're rising, where is that excess money going?

Hope that helps!

It definetly helps, yes. But let me rephrase your sentence:

"But ask yourself this: if assets in your portfolio are falling less than they're rising, where is that excess money coming?"

Those two sentences seem to make sense evenly.

I suppose that the problem is complex, I don't have an economics background, but there is no "constancy law" of money, QE's in US, Draghi's bazooka in Europe and Abe in Japan are making the money mass to increase, but those matters are well beyond my "head". I am unlikely to produce useful ideas here.

Is it just kind of "the general state of the world economy" what is going on here, meaning that whatever it be?

[Pointedstick]

It definetly helps, yes. But let me rephrase your sentence:

"But ask yourself this: if assets in your portfolio are falling less than they're rising, where is that excess money coming?"

Those two sentences seem to make sense evenly.

It is being created from thin air, either from government money printing or bank lending. Economic growth without an at least equal rise in the quantity of money leads to deflation, which will cause asset prices to fall in the aggregate, not rise. If asset prices are falling, either there's not enough money in the economy, or the assets were overvalued to begin with and should eventually settle at a more accurate price level to better reflect their more correct valuation.

[aaa]

I see this community is REALLY responsive, I didn't see this replay.

Each of the assets has a non-negative expected real return.  According to the numbers posted at http://www.bogleheads.org/wiki/Historic ... ed_returns the expected real returns are:

Cash (0-2 year Treasuries): 0.3%
Long Term (20 year) Treasuries: 1.5%
Total stock market: 5%
Gold: 0% (Bogleheads in general dislike gold, so don't list an expected return, but I think 0% is the general assumption for gold's expected return)

By real return I meant "inflation adjusted". Anyway, the idea is that those assets have expected positive returns and by building a portfolio, mutual correlations reduce variability, while the expected return is an average of the individual expected returns.

If we combine these by adding them and dividing by 4, we get only 1.7% - but that's not the right way to determine an expected outcome from a collection of assets.  A better way is to use Modern Portfolio Theory (MPT, see http://en.wikipedia.org/wiki/Modern_portfolio_theory) which uses the expected return, standard deviation, and correlation of individual assets to determine the expected return and deviation of a collection of assets in some proportion.  I don't know the exact real return predicted by MPT for the PP, by I'd guess it's in the 2% range.  This number includes the benefit of rebalancing.

Yes, the theory you refer to (it's called in efect "modern" but dates back to Markowitz in the 70's) predicts that the expected return of a single year (prior to rebalancing) woud be the sum divided by 4 (in that same wikipedia entry the formula is "Expected return" inside a big rectangle full of formulas).

Here's where Browne's genius comes into play.  MPT is based on some assumptions that don't actually turn out to be true - specifically that investment returns in a particular asset are random following a normal distribution curve and correlations between assets are fairly consistent over time.  In reality, investment returns are not predictable (but not random following a normal distribution curve) and correlations between assets can (and do) change dramatically.  The assets comprising the PP are chosen because they act in predictable (not random) ways in various economic conditions.  The "extra" 1% comes from this predictability.  Gold tends to go up in response to fears of inflation.  Long term bonds tend to go up during periods of deflation.  Stock tends to up in periods of prosperity.  I don't see any of these changing any time soon.

This makes a lot of sense. I am reading now a book by Benoit Mandelbrot (The (mis)behavior of markets) in which he talks about the non-normality (fat tails) of the financial asset returns, something that seems very well established. He also talks about the impredictability of the markets but he says there are kind of "hidden correlations" (multifractality) on the price assets, that make them not independent, as modern portfolio theory requieres. Everything quite misterious, I am still reading the book, so I cannot say more.

MediumTex is J.M.Lawson.  Craigr (who also posts here) is Craig Rowland.

I take note.

Thank you.

[Tortoise]

aaa, ignoring historical returns, this is the simplest way I can think of to explain why the PP tends to produce real returns over time:

1. Real Stock Returns: A result of economic expansion driven by human ingenuity and foresight (entrepreneurship, innovation, and business investment).

2. Real Bond Returns: A result of borrowers making interest payments to bondholders. It's simply a transfer of wealth from debtors to creditors. (See Time Value of Money.)

3. Rebalancing Gains: The price volatility of the component assets means that rebalancing tends to result in buying low and selling high. You are capturing or "harvesting" volatility and converting it into a return. Effectively, the market is "paying" you in return for making the asset prices a bit more stable for everyone else.

None of those reasons is unique to the PP. A conventional stock-and-bond portfolio will tend to produce real returns over time for the same reasons. The unique benefits of the PP come from its wider diversification and higher security as described elsewhere.

[MediumTex]

Think of the three volatile PP assets as unusual springs similar to the kind you see on automobile suspensions. 

These springs can be compressed and expanded, but they can only be compressed so much, and the more you compress them, the harder it becomes to compress them more.

When it comes to the springs' expansion, however, something interesting happens, and we find that the springs can be stretched to almost absurd lengths, and once the springs are stretched in this way, that becomes their new resting position and compressing them from that point becomes just as hard as it was at the earlier point before they were stretched.

If you have this spring analogy straight in your head, consider how it applies to the PP:

Gold went from under $100 an ounce in the early 1970s to $800 by the end of the decade.  From there, gold went on to have a terrible bear market that lasted two decades, but the lowest gold ever went during this bear market was still over triple the price it started at in the early 1970s.  The spring had a new resting position at around $350-$400 an ounce during this period, MUCH higher than the resting position it started at in the early 1970s.

The S&P 500 went from just over 100 in 1981 to 1500 in 2000.  From there, stocks entered a bear market and the stock market "collapsed" to the 600s in 2009.  Note, however, that, like gold, its new low point was still 6 times higher than the previous low point.  The stock "spring" had new characteristics after having been stretched over the 20 year bull market.

In a nutshell, when the PP's assets do poorly they tend to lose 40-50% of their value (sometimes more, but it's rare).  However, when the PP's assets do well, they routinely see gains of several hundred percent, easily wiping out the losses in the other assets in the portfolio.

Day to day, you don't see this dynamic, just like you don't see a tree growing, but that's what is happening within the portfolio.

I hope that helps.

[brownehead]

rickb gave a very good explanation and MediumTex's is intuitive too. However, there is still the question about how the hell Harry Browne made his calcs for the estimated long term Permanent Portfolio real returns. In fact, I think in his first books (3 decades ago) he estimated 5% real return, and 1972-2012 real returns for the PP are just that! (~9% nominal returns and ~4% official inflation).

[dragoncar]

Each of the assets has a non-negative expected real return.  According to the numbers posted at http://www.bogleheads.org/wiki/Historic ... ed_returns the expected real returns are:

Cash (0-2 year Treasuries): 0.3%
Long Term (20 year) Treasuries: 1.5%
Total stock market: 5%
Gold: 0% (Bogleheads in general dislike gold, so don't list an expected return, but I think 0% is the general assumption for gold's expected return)

If we combine these by adding them and dividing by 4, we get only 1.7% - but that's not the right way to determine an expected outcome from a collection of assets.  A better way is to use Modern Portfolio Theory (MPT, see http://en.wikipedia.org/wiki/Modern_portfolio_theory) which uses the expected return, standard deviation, and correlation of individual assets to determine the expected return and deviation of a collection of assets in some proportion.  I don't know the exact real return predicted by MPT for the PP, by I'd guess it's in the 2% range.  This number includes the benefit of rebalancing.

So, where does the "extra" 1% come from?

Here's where Browne's genius comes into play.  MPT is based on some assumptions that don't actually turn out to be true - specifically that investment returns in a particular asset are random following a normal distribution curve and correlations between assets are fairly consistent over time.  In reality, investment returns are not predictable (but not random following a normal distribution curve) and correlations between assets can (and do) change dramatically.  The assets comprising the PP are chosen because they act in predictable (not random) ways in various economic conditions.  The "extra" 1% comes from this predictability.  Gold tends to go up in response to fears of inflation.  Long term bonds tend to go up during periods of deflation.  Stock tends to up in periods of prosperity.  I don't see any of these changing any time soon.

Medium Tex, by what you say "...our book...", I understand that you are one of the authors, Roland&Lawson, of the PP book. I consider a privilege to be able to talk about this topic with you.

MediumTex is J.M.Lawson.  Craigr (who also posts here) is Craig Rowland.

I buy everything you said, except for the "extra 1% comes from predictability" part.  Basically you are saying we should expect a 2% real return, plus 1% from predictability = 3% real? 

The 2% part makes sense.  But can anyone elaborate on why we should expect that extra 1% (recall that its not due to rebalancing... Or is it that rebalancing is more effective because of the assets we gold?)

I must say I'm really glad nobody has trotted out the "it just works" explanation I've see so many times before.  Thinking of the PP like a floating box, or a race at doesn't really help me.  Medium tex, I appreciate the effort, but your spring example seems to mostly demonstrate that on average the assets will rise over time.  But not how we beat the 2% expected real return (and yes I understand the asymmetry between the spring expansion and compression).

[Pointedstick]

Personally, I put the rebalancing premium at an additional percentage point or more beyond what MPT would assume, because of two things: the greater volatility of the rebalanced assets, and the greater time before rebalancing by using discrete and wide bands rather than blindly rebalancing yearly. This has driven my quest to increase the PP's volatility so as to extract even more juice out of the rebalancing premium. Widening the bands has been backtested to juice those returns a bit as well, I believe.

[Mdraf]

Personally, I put the rebalancing premium at an additional percentage point or more beyond what MPT would assume, because of two things: the greater volatility of the rebalanced assets, and the greater time before rebalancing by using discrete and wide bands rather than blindly rebalancing yearly. This has driven my quest to increase the PP's volatility so as to extract even more juice out of the rebalancing premium. Widening the bands has been backtested to juice those returns a bit as well, I believe.

Doesn't widening the bands imply you need a longer time horizon? (ie you have to be young)

[Pointedstick]

Doesn't widening the bands imply you need a longer time horizon? (ie you have to be young)

That's a good question. It would increase the momentum effect, I guess potentially lengthening the amount of time between rebalances, and potentially failing to capture smaller rises and falls in the asset prices. I would imagine that all of this would increase the volatility, so yeah, I guess you'd need a longer time horizon. Maybe not by that much though.

[Mdraf]

Doesn't widening the bands imply you need a longer time horizon? (ie you have to be young)

That's a good question. It would increase the momentum effect, I guess potentially lengthening the amount of time between rebalances, and potentially failing to capture smaller rises and falls in the asset prices. I would imagine that all of this would increase the volatility, so yeah, I guess you'd need a longer time horizon. Maybe not by that much though.

That's my main concern with the PP.  One needs  a 30-40 year working career ahead to feel very comfortable

[Pointedstick]

That's my main concern with the PP.  One needs  a 30-40 year working career ahead to feel very comfortable

Funny, I feel the opposite. A long time horizon is necessary for smoothing portfolio volatility across time, right? If so, then wouldn't that impy that the PP--a much less volatile portfolio than nearly any other that's not extremely cash-heavy--would require a shorter time horizon?

I'd want a 30-40 year working career if I was like 60% stocks or more. The PP makes me feel much more comfortable with its lower yearly drawdowns compared to something like that.

[Mdraf]

That's my main concern with the PP.  One needs  a 30-40 year working career ahead to feel very comfortable

Funny, I feel the opposite. A long time horizon is necessary for smoothing portfolio volatility across time, right? If so, then wouldn't that impy that the PP--a much less volatile portfolio than nearly any other that's not extremely cash-heavy--would require a shorter time horizon?

I'd want a 30-40 year working career if I was like 60% stocks or more. The PP makes me feel much more comfortable with its lower yearly drawdowns compared to something like that.

Good point

[rickb]

I buy everything you said, except for the "extra 1% comes from predictability" part.  Basically you are saying we should expect a 2% real return, plus 1% from predictability = 3% real? 

The 2% part makes sense.  But can anyone elaborate on why we should expect that extra 1% (recall that its not due to rebalancing... Or is it that rebalancing is more effective because of the assets we gold?)

What I'm saying is that the "expected" return is based on a mathematical model that doesn't match reality - specifically returns are not randomly distributed with a normal distribution and with persistent correlations - so the MPT "expected" return isn't what you should actually expect.  In the model, the assets go up and down randomly, with a specific standard deviation around a specific average with a specific correlation to each other.  For example, the model says long term bonds and stocks have a slight positive correlation (0.127 according to http://www2.stetson.edu/fsr/abstracts2/V14-3%20A5.pdf), even during periods of inflation with rising interest rates (!!??).  If you only look at periods of inflation with rising interest rates I'd be surprised if the actual correlation between long term bonds and stocks isn't something pretty close to -1 (and the average return of long term bonds during such a period sure as hell isn't going to be 5.8% nominal).

I don't know whether Browne did this (my guess is no, since he didn't have a period of deflation to use as a data point), but a more nuanced way to arrive at an expected long term return would be to divide history into the overall market conditions the portfolio is designed to cover (inflation, deflation, prosperity, and "tight money" recession) and compute average returns, standard deviations, and correlations for each of the assets within each of these conditions, and then compute an MPT expected return in each condition.  The long term return is then something like the sum of the products of the returns during each of these market conditions times the probability of the market condition occurring.

The overall point here is that the mathematical model ignores the prevailing market conditions (which can in general only be determined in hindsight) and treats the individual assets as if their relative returns can never be more closely predicted than their long term average, standard deviation, and correlations would indicate.  But in reality, the PP assets act predictably in various economic conditions, and this predictability invalidates the conclusions the model produces.  If you want a more accurate prediction, you need to use a more complicated model.

[glennds]

Doesn't widening the bands imply you need a longer time horizon? (ie you have to be young)

That's a good question. It would increase the momentum effect, I guess potentially lengthening the amount of time between rebalances, and potentially failing to capture smaller rises and falls in the asset prices. I would imagine that all of this would increase the volatility, so yeah, I guess you'd need a longer time horizon. Maybe not by that much though.

I speculate as to whether the increased momentum effect of wider bands would also bring increased risk with it. It seems that the idea is to keep the portfolio reasonably balanced because (i) none of us can predict which of the four asset classes will fall into favor and when, and (ii) when the shift happens, the newly favored asset will need to be sufficiently weighted to carry the portfolio. In other words, if the bands are really widened, in effect you're favoring an asset class materially more than another, probably for an extended period of time, and I'm not sure it's still the PP doctrine at that point.

[frommi]

The PP worked in the past because only very few people where using it. Nowadays everybody is using asset allocation strategies in which bonds and cash are a greater part. This drives down returns for all AA strategies, simply because there is no return above the risk-free-rate without risk. Just try to imagine what would happen if everybody uses the PP or an equivalent AA strategy. All assets would go coupled in lockstep as new money is funded or withdrawn. The PP worked in the past but it is not guaranteed to work in the future. So even a PP user is speculating.
To get really wealthy you have to use your brain and learn learn learn. ( and buy productive assets cheap )

[Kshartle]

If the idea is that you make more in bull markets than bear markets..........and rebalancing occurs......way more frequently than bull markets start and end.......wider bands are going to be way better return-wise.

And while they might be more volitile in the short-run, at some point not far off they've got to 99% meet at a minimum no matter how bad it went.

Has anyone tested in for 60 -10/5 before or something ridiculous? Really letting a bull market run.

[Kshartle]

The PP worked in the past because only very few people where using it. Nowadays everybody is using asset allocation strategies in which bonds and cash are a greater part. This drives down returns for all AA strategies, simply because there is no return above the risk-free-rate without risk. Just try to imagine what would happen if everybody uses the PP or an equivalent AA strategy. All assets would go coupled in lockstep as new money is funded or withdrawn. The PP worked in the past but it is not guaranteed to work in the future. So even a PP user is speculating.
To get really wealthy you have to use your brain and learn learn learn. ( and buy productive assets cheap )

Wait. I thought I could get rich ignoring everything around me, pretending it was impossible to make any reasonabl prediction about the future and still do as well the most volitile asset class with half the risk?  kidding....

What the PP does is bet that purchasing power will trend between these different assent classes at, fairly close to even intervals. And it's pretty much always going to be one of these based on the world currently. By owning all four, and we can almost cut out cash.......you are certain to own whatever goes on a bull run and let it run for a while.

The 3 big ones don't get out of whack that often, especially if you're adding so they can run a while.

I think if you're younger the PP is a little foolish because you can easliy avoid cash and bonds for a long time at these high of prices.

The PP will work in the long run. But I think it's in for a big crash. The sooner it is the smaller it will be. If they hold off for a couple years it will be pretty big. It might not be nominal but it will certainly be real.

I personally will not buy in until after the crash. I think that will be much higher dividends, lower PEs, higher gold price and lower bond price.