The Futility of Predicting Future Returns

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Maynard F. Speer
Posts: 2139
Joined: Wed Mar 18, 2015 10:31 am

Re: The Futility of Predicting Future Returns

Post by Maynard F. Speer »

HomerJ wrote:Citation please... ALL value stocks or value stock funds performed this way? Or, looking back, you've found a couple of value stock funds (out of a hundred) that did this well during the dot-com boom?
Take your pick .. If you'd been in value stocks you wouldn't have been exposed to overvalued tech stocks in the first place

Image

If you use this, you are market timing.

I don't agree with you jumping from one market to another based on valuations, but I understand it... What I don't understand is you claiming you're not market timing. You believe now is a bad time to be in one market based on valuations, but a good time to be in another market based on valuations.

Moving into a market at one time, and out of a market at another time based on some "signal" is basically the DEFINITION of market-timing.
You don't have to jump between markets .. It can be primarily a purchasing principle .. e.g. Warren Buffett's the world's most famous value investor, and for him that just means buying at good value ..

Anyone who's rebalanced a portfolio has probably done more selling and rotating out of assets than Buffett has .. So it's up to you how you use measures of market value
"Economics is a method rather than a doctrine, an apparatus of the mind, a technique of thinking, which helps its possessor to draw correct conclusions." - John Maynard Keynes
User avatar
siamond
Posts: 5792
Joined: Mon May 28, 2012 5:50 am

Re: The Futility of Predicting Future Returns

Post by siamond »

As I said... It quickly turns in a religious debate (notably with the tone set by the OP). Which means that the only exercise in futility here is to even try to discuss the matter in a rational manner. Hopeless. Both sides will stick to their guns, and zero progress will be made.

Personally, I would just point out that two of our most respected 'illuminati' (John Bogle and W. Bernstein) have written countless pieces documenting their latest expected returns estimates (and the corresponding math), to drive various points and suggesting subsequent policy decisions for various interested parties (individual investors, pension funds, retirement funds, name it). Rick Ferri regularly updates his 30yrs forecast. Etc. Why would those amazing people bother if there was not a kernel of (actionable) truth in there? And they are all being extremely clear that this is a statistical analysis with all its caveats (like most things in investing anyway), and yet it is indeed useful to those who choose to listen with an open mind...

And I'll leave it at that, because it's futile to even try to establish the necessary shades of grey here... :|
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

siamond wrote:As I said... It quickly turns in a religious debate (notably with the tone set by the OP). Which means that the only exercise in futility here is to even try to discuss the matter in a rational manner. Hopeless. Both sides will stick to their guns, and zero progress will be made.

Personally, I would just point out that two of our most respected 'illuminati' (John Bogle and W. Bernstein) have written countless pieces documenting their latest expected returns estimates (and the corresponding math), to drive various points and suggesting subsequent policy decisions for various interested parties (individual investors, pension funds, retirement funds, name it). Rick Ferri regularly updates his 30yrs forecast. Etc. Why would those amazing people bother if there was not a kernel of (actionable) truth in there? And they are all being extremely clear that this is a statistical analysis with all its caveats (like most things in investing anyway), and yet it is indeed useful to those who choose to listen with an open mind...

And I'll leave it at that, because it's futile to even try to establish the necessary shades of grey here... :|
Siamond,

Let's avoid labeling each other or our tones.

Given that you believe in valuations, tell me. As of today, which of U.S. stocks or U.S. bonds have the highest expected forward 5-year risk-adjusted returns? How about 10 years? 20? 30? Don't forget to provide the ranges, for each of your predictions.

By the way, back in the day, many, many years ago, when I took a probability course, an "Expected Value" was just the result of a calculation over a set of data. The meaning of "expected" was, if applied to historical data, that it would have been the thing to expect before the data unrolled. So, an expected value based on past data predicts the past! If you are to provide a mathematical future "expected" value, you'll have to compute it on future data, of course. ;)

Can you provide (1) tight-enough ranges and (2) guarantees, so that your predictions would be useful to me?

Otherwise, what was false in my OP?

Let me rehash its important points so that you can criticize them specifically:
  1. I said that I have learned that by living below my means and investing in Total Market index funds, I will reap my fair share of the future returns of three broad markets.
  2. I said that I just needed to select a reasonable ratio of bonds and stocks, and rebalance my portfolio once a year; that this lowered my anxiety about selecting the best asset allocation.
  3. I said that to make an asset allocation choice, there is no need for taking into consideration future return predictions. Note that I did not say that one cannot or should not take them into consideration; I said that one does not need to do so.
  4. I said that I found liberating to toss future return predictions aside and accept market returns, as uncertain as they are; that I wanted to encourage all Bogleheads to do it.
  5. Finally, I provided a proof that the future returns of broad bond and stock funds cannot be predicted with enough accuracy to predict the highest returning asset over a selected period of time, in the context of people using various metrics to predict which asset class has the highest future return to invest most of their money into it. (I'm not talking of mild tilts, here).
So, what's wrong in what I wrote? Can you provide better actionable advice?
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
2015
Posts: 2906
Joined: Mon Feb 10, 2014 2:32 pm

Re: The Futility of Predicting Future Returns

Post by 2015 »

This is one of the best threads I've read on this site in a long time. So much better than deciding which refrigerator to buy (although valuable in its own right!). longinvest, thanks for starting the thread and for your very insightful posts.

As to the question posed above re why are the likes of Bernstein, Bogle, Ferri, et al provide estimates of future returns (when all the literature out there points to the high degree of inaccuracy of such)? Because the public craves certainty, real or imagined. It's called giving people what they want and lord, how people want the soothsayer to look into that crystal ball.
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

I'll let others try to address the other issues if they wish, but just so we're on the same page...
longinvest wrote:By the way, back in the day, many, many years ago, when I took a probability course, an "Expected Value" was just the result of a calculation over a set of data. The meaning of "expected" was, if applied to historical data, that it would have been the thing to expect before the data unrolled. So, an expected value based on past data predicts the past! If you are to provide a mathematical future "expected" value, you'll have to compute it on future data, of course. ;)
Calculating the mean of a data set is not the expected value. Depending on how you do sampling, what the underlying distribution is, etc., it may or may not be a decent estimate of the expected value.

We're dealing with stochastic processes with statistical characteristics that probably change over time. As you say, we are looking to the future, not the past. We are talking about what we think future distributions will more likely be. Expressing a view on what we think (future) expected value will be does not require crunching future data, as stated above. It is a guess, a prediction.

It probably will in part be informed by historical data, among other factors. Some people believe that the best estimate of future values is realistically determined by taking the mean of the past data. (This is better and more the case the better the historical mean reflects the actual expected value of the underlying distribution, and the more the future looks like the past.) The ones talking valuations, dividend yields, and other metrics are not in this camp—they believe some other information can improve the estimate over just taking the historical mean, that markets change and furthermore some statistics capture some of the changes at some level better than pure chance.

It's important to note that the actual expected value is unknown, even in hindsight. Because of the randomness, the observed outcomes may be biased relative to the underlying distributions. We never know if a certain past return of 8% happened because for example the expected value was 6% and we got lucky, or the expected value was 10% and we happened to get unlucky, or anything else. Furthermore, many think that the distributions are likely changing over time (were the distribution stationary, we would assume over the long run that we are not getting consistently lucky or consistently unlucky, and thus come into superior estimates as time goes on over what the statistical properties are, with more data). This all implies a very high level of uncertainty in all these forecasts, making static portfolio allocations a very defensible choice.

But without even basic long-term assumptions like "I think stocks will return more than cash" there's no way to make any real decisions about asset allocation other than pulling figures out of a hat or throwing darts. You have to start from something, even if it's just simple heuristics. Relying on heuristics may in fact do better long term than crunching the numbers and reallocating a lot; predicting which method will work better in advance is a statement about how we view the markets and how they work. In real life, statistics is always subjective to some degree.
User avatar
Ketawa
Posts: 2340
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: The Futility of Predicting Future Returns

Post by Ketawa »

longinvest wrote:Given that you believe in valuations, tell me. As of today, which of U.S. stocks or U.S. bonds have the highest expected forward 5-year risk-adjusted returns? How about 10 years? 20? 30? Don't forget to provide the ranges, for each of your predictions.
You are putting words in the mouths of people who think valuations are meaningful. I'm pretty sure those who are citing valuations in this thread have said nothing about risk-adjusted returns. U.S. stocks have higher expected returns, obviously. Valuations are useful for creating an estimate of expected returns, and Simplegift has posted a lot of charts showing how. You have just ignored them.
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

There also seems to be some changing of the language, or I just misinterpreted some things. It's hard to get what is meant.
longinvest wrote:[*] I said that to make an asset allocation choice, there is no need for taking into consideration future return predictions. Note that I did not say that one cannot or should not take them into consideration; I said that one does not need to do so.
First of all, the thread title is "The Futility of Predicting Future Returns." That implies one should not take them into consideration, unless they like exercises in futility. (I'm pretty sure, that by responding here, I don't have enough of an aversion to such exercises. :shock: )
longinvest wrote:It is so liberating to toss future return predictions aside and simply embrace accepting market returns, as uncertain as they are. I want to encourage all Bogleheads to do it.
Sounds like telling people to toss future return predictions aside. Presumably, before use and not after.

Fundamentally, going back to this:
longinvest wrote:Can you provide (1) tight-enough ranges and (2) guarantees, so that your predictions would be useful to me?
It depends on what kind of edge or information you think is actionable and useful.

Let's say you're a sports bettor and you're betting on the outcome of a game (versus the points spread). If you assume the line is placed at the median outcome (in practice this is not a great example because it's frequently not), there's a 50/50 chance of winning the bet. Let's say you do some snooping around and notice that the star player on one team suddenly came down with flu-like symptoms but nobody's reported on it yet. Is this information actionable?

We still obviously can't guarantee who will win, but the odds have improved, perhaps enough in your favor that you might be expected to earn money by betting on the opposite team, even net of fees. Different people may disagree about what the expected value of a bet on the other team might be, what the information really means. It depends. Context is big.

Even in a market where you don't have unique information, if there is some feature about the conditions and setup—e.g. people seem to bet too frequently on speculative small growth stocks with zero profitability, resulting in low risk-adjusted returns historically—this may translate to an actionable edge if you suspect that certain features seen in the past have some chance of continuing to some degree in the future. No guarantees, but maybe you're doing better than flipping a coin now. These things are based on return expectations, that certain stocks will do worse than other ones, even. Never mind the much less controversial positions of "stocks have higher expected returns than bonds, probably."
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

lack_ey wrote:There also seems to be some changing of the language, or I just misinterpreted some things. It's hard to get what is meant.
longinvest wrote:[*] I said that to make an asset allocation choice, there is no need for taking into consideration future return predictions. Note that I did not say that one cannot or should not take them into consideration; I said that one does not need to do so.
First of all, the thread title is "The Futility of Predicting Future Returns." That implies one should not take them into consideration, unless they like exercises in futility. (I'm pretty sure, that by responding here, I don't have enough of an aversion to such exercises. :shock: )
I think that it is futile to use predictions, because one has no guarantee to select the best asset allocation that will deliver the highest return. One might as well loose all the anxiety and select a good enough asset allocation using a Three-Fund Portfolio.

There is no contradiction, here. Just because I think that it is futile, doe not mean that one cannot make or use predictions to select an asset allocation (AA); one can, but I think that it is futile. If one cannot sleep at night without using predictions to select an AA, then one should use predictions to set his AA; but this will remain futile, because one will not get an outperformance guarantee out of using these predictions. But, worse, if the predictions are used to put all one's money into one asset, then this could be harmful, not only futile.


I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
lack_ey wrote:Fundamentally, going back to this:
longinvest wrote:Can you provide (1) tight-enough ranges and (2) guarantees, so that youingr predictions would be useful to me?
It depends on what kind of edge or information you think is actionable and useful.

Let's say you're a sports bettor and you're betting on the outcome of a game (versus the points spread). If you assume the line is placed at the median outcome (in practice this is not a great example because it's frequently not), there's a 50/50 chance of winning the bet. Let's say you do some snooping around and notice that the star player on one team suddenly came down with flu-like symptoms but nobody's reported on it yet. Is this information actionable?
You said, yourself, in your explanation of "expected value", that:

"Expressing a view on what we think (future) expected value will be does not require crunching future data, as stated above. It is a guess, a prediction."

and

"In real life, statistics is always subjective to some degree."

So, there is an embedded risk in believing a future return prediction, because the prediction is necessarily tainted by subjective choices of the person making the prediction.

If you consider, on top of this prediction risk, the large error range of the prediction (which does not account for the subjectivity risk), then you end up into arbitrary territory; you must acknowledge that you are making a bet. You've simply chosen a different approach than tossing a coin.

The framework I explained (that I have learned on this forum) to put no less than 25% in each of bonds and stocks, harvest market returns (guaranteed) in three markets, while considering that bonds are less volatile than stocks, but that stocks can deliver much higher (or lower) returns, is hard to improve upon.

Note that this framework does not in any way depends on analyzing past return data. Not betting more than 3:1 on one asset is just rational. That bonds (total market, of course) have limited volatility is just due to mathematics of bond funds. That stocks have a chance at much higher growth (but can be subject to crashes, too), is due to the nature of stocks. You can analyze past data across all countries; this will always be true, whether stocks or bonds outperform.
We still obviously can't guarantee who will win, but the odds have improved, perhaps enough in your favor that you might be expected to earn money by betting on the opposite team, even net of fees. Different people may disagree about what the expected value of a bet on the other team might be, what the information really means. It depends. Context is big.
This defies basic logic. This would mean that, lucky me longinvest, I have been privy to some secret information available publicly on the Bogleheads forum in posts of generous benefactors of market-beating advice.

Why should I believe that publicly available information hasn't been acted upon by institutional investors? Why would all the hedge funds, pension funds, and others ignore Fama-French research on factor investing? Why should only the select few Bogleheads members, that have been so lucky to be exposed to this theory daily in the forums, be the only ones acting on this information?

If, instead, institutional money has already acted on this information; why should I believe that market prices have not already adjusted? Why shouldn't I consider the possibility that I'm be the goat to be sacrificed, here (and maybe you too)?

:oops:
lack_ey wrote:Even in a market where you don't have unique information, if there is some feature about the conditions and setup—e.g. people seem to bet too frequently on speculative small growth stocks with zero profitability, resulting in low risk-adjusted returns historically—this may translate to an actionable edge if you suspect that certain features seen in the past have some chance of continuing to some degree in the future. No guarantees, but maybe you're doing better than flipping a coin now. These things are based on return expectations, that certain stocks will do worse than other ones, even. Never mind the much less controversial positions of "stocks have higher expected returns than bonds, probably."
Demand and supply does not rule out the possibility that a segment of the market is priced, often enough, so as to deliver higher returns, if it exposes investors to higher volatility and risk. But, I do not believe in free lunches.

In other words, one can always get a chance at even higher returns by taking the risk of even lower returns. There are many ways to do that, including tilting and leverage.

I am relatively conservative. I prefer to be the tortoise than the hare, even if it's not flashy.

I do believe in mathematics, though. So, I know that when I invest in a low-cost total-market index fund, I am guaranteed to reap market returns, whatever they are. Not only that, but I'll also beat the active investors of that market in the aggregate, after fees. (Sharpe's theorem).
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
User avatar
JoMoney
Posts: 11353
Joined: Tue Jul 23, 2013 5:31 am

Re: The Futility of Predicting Future Returns

Post by JoMoney »

lack_ey wrote:...Let's say you're a sports bettor and you're betting on the outcome of a game (versus the points spread). If you assume the line is placed at the median outcome (in practice this is not a great example because it's frequently not), there's a 50/50 chance of winning the bet. Let's say you do some snooping around and notice that the star player on one team suddenly came down with flu-like symptoms but nobody's reported on it yet. Is this information actionable?

We still obviously can't guarantee who will win, but the odds have improved, perhaps enough in your favor that you might be expected to earn money by betting on the opposite team, even net of fees. Different people may disagree about what the expected value of a bet on the other team might be, what the information really means. It depends. Context is big.

Even in a market where you don't have unique information, if there is some feature about the conditions and setup—e.g. people seem to bet too frequently on speculative small growth stocks with zero profitability, resulting in low risk-adjusted returns historically—this may translate to an actionable edge if you suspect that certain features seen in the past have some chance of continuing to some degree in the future. No guarantees, but maybe you're doing better than flipping a coin now. These things are based on return expectations, that certain stocks will do worse than other ones, even. Never mind the much less controversial positions of "stocks have higher expected returns than bonds, probably."
I think this a reasonable analogy. A 'tilt' isn't much more then a gamble that one side is going to do better than another. The problem is, if it's not based on the idea of a 'risk premium', where there are sufficient people on the other side of the bet who willingly pay more to avoid the 'risk' of that style, then it's simply a behavioral thing and it's just as likely that that the performance chasers that previously bet on one side might switch to the other if it's believed that's where superior returns are.
We can prove excess market returns is a zero-sum game. I'm willing to accept there are patterns of repeated bad behavior among investors, the difference is I think the odds suggest most of us are more likely to be making a mistake by picking a side. The information available suggesting that some mutual fund or active style is inherently 'better' than another is not unique information, and very much resembles every other story designed to appeal to those who want to believe they're 'above average'.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham
BahamaMan
Posts: 896
Joined: Wed Oct 01, 2014 5:52 pm

Re: The Futility of Predicting Future Returns

Post by BahamaMan »

longinvest wrote: I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
Exactly ! :thumbsup ......

I have been amused at 'Value Investors' or 'Dividend Stock Investors' over the last 20 years. They are just a lot smarter than index investors. They remind me of the people I talk to that go to Las Vegas. They always win errrrr.......They always tell you when they win. :mrgreen:
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

longinvest wrote:
lack_ey wrote:There also seems to be some changing of the language, or I just misinterpreted some things. It's hard to get what is meant.
longinvest wrote:[*] I said that to make an asset allocation choice, there is no need for taking into consideration future return predictions. Note that I did not say that one cannot or should not take them into consideration; I said that one does not need to do so.
First of all, the thread title is "The Futility of Predicting Future Returns." That implies one should not take them into consideration, unless they like exercises in futility. (I'm pretty sure, that by responding here, I don't have enough of an aversion to such exercises. :shock: )
I think that it is futile to use predictions, because one has no guarantee to select the best asset allocation that will deliver the highest return. One might as well loose all the anxiety and select a good enough asset allocation using a Three-Fund Portfolio.

There is no contradiction, here. Just because I think that it is futile, doe not mean that one cannot make or use predictions to select an asset allocation (AA); one can, but I think that it is futile. If one cannot sleep at night without using predictions to select an AA, then one should use predictions to set his AA; but this will remain futile, because one will not get an outperformance guarantee out of using these predictions. But, worse, if the predictions are used to put all one's money into one asset, then this could be harmful, not only futile.


I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
Okay, that is fair enough. I dropped a couple words here and there too, so I can't say anything. I just wanted to make sure I got what you were after.

By the way, I don't know of anybody here seriously suggesting to "put all one's money into one asset" based on any kind of predictions, so this seems an irrelevant straw man. Even if somebody knew the actual expected values, covariances, etc. (never mind the overconfidently estimated ones), this is probably a bad idea, especially if we are looking at relatively narrow sub-asset classes. The risk-adjusted return of such a bet is a priori low. You still always want to diversify, and if anybody is making any change at all based on current market conditions, it should obviously be with the understanding of the low level of predictability of many parts of the markets and the fact that one's predictions could be varying degrees of wrong.
longinvest wrote:
lack_ey wrote:Fundamentally, going back to this:
longinvest wrote:Can you provide (1) tight-enough ranges and (2) guarantees, so that youingr predictions would be useful to me?
It depends on what kind of edge or information you think is actionable and useful.

Let's say you're a sports bettor and you're betting on the outcome of a game (versus the points spread). If you assume the line is placed at the median outcome (in practice this is not a great example because it's frequently not), there's a 50/50 chance of winning the bet. Let's say you do some snooping around and notice that the star player on one team suddenly came down with flu-like symptoms but nobody's reported on it yet. Is this information actionable?
You said, yourself, in your explanation of "expected value", that:

"Expressing a view on what we think (future) expected value will be does not require crunching future data, as stated above. It is a guess, a prediction."

and

"In real life, statistics is always subjective to some degree."

So, there is an embedded risk in believing a future return prediction, because the prediction is necessarily tainted by subjective choices of the person making the prediction.

If you consider, on top of this prediction risk, the large error range of the prediction (which does not account for the subjectivity risk), then you end up into arbitrary territory; you must acknowledge that you are making a bet. You've simply chosen a different approach than tossing a coin.

The framework I explained (that I have learned on this forum) to put no less than 25% in each of bonds and stocks, harvest market returns (guaranteed) in three markets, while considering that bonds are less volatile than stocks, but that stocks can deliver much higher (or lower) returns, is hard to improve upon.

Note that this framework does not in any way depends on analyzing past return data. Not betting more than 1:3 on one asset is just rational. That bonds (total market, of course) have limited volatility is just due to mathematics of bond funds. That stocks have a chance at much higher growth (but can be subject to crashes, too), is due to the nature of stocks. You can analyze past data across all countries; this will always be true, whether stocks or bonds outperform.
It seems that you completely ignored the example and started segueing into something entirely different.

Let's simplify it even more. Let's say there's an investment where every day somebody flips a coin. Heads, you double your money. Tails, you lose it all. If I tell you that heads comes up 55% of the time because the coin is weighted that way, is that actionable information (assuming it is true)? If you don't believe it is, then everything makes more sense. If you do believe it is, then there's more explaining to be done.

Most everybody putting money in stocks or other risky assets is making a bet, one that they hope will pay off. Some don't place bets on gold (arguable) or the good-for-nothing brother-in-law's get-rich-quick scheme because they implicitly or explicitly think the risk/return characteristics are worse. This doesn't necessarily require explicit estimations of actual returns.

You've explained your framework, but without any expectations around the characteristics of different asset classes, one might wonder why stocks and bonds—but not other things—are chosen. And, as I've asked before, why take the extra risk on stocks up to 75% stocks without an expectation of higher returns?
longinvest wrote:
We still obviously can't guarantee who will win, but the odds have improved, perhaps enough in your favor that you might be expected to earn money by betting on the opposite team, even net of fees. Different people may disagree about what the expected value of a bet on the other team might be, what the information really means. It depends. Context is big.
This defies basic logic. This would mean that, lucky me longinvest, I have been privy to some secret information available publicly on the Bogleheads forum in posts of generous benefactors of market-beating advice.

Why should I believe that publicly available information hasn't been acted upon by institutional investors? Why would all the hedge funds, pension funds, and others ignore Fama-French research on factor investing? Why should only the few select few Bogleheads members, that have been so lucky to be exposed to this theory daily in the forums, be the only ones acting on this information?

If, instead, institutional money has already acted on this information; why should I believe that market prices have not already adjusted? Why shouldn't I consider the possibility that I'm be the goat to be sacrificed, here (and maybe you too)?

:oops:
Again it's unclear what you're responding to, as it doesn't seem to be the example given. In the example, there is legitimately unique information, but now you're going back to questioning uniqueness of information in financial markets?

The whole factor investing tirade is probably a topic for another thread, as that's not even necessary here. [But some may want to consider (1) limits and costs of arbitrage, including leverage aversion, (2) institutional money manager incentives, in terms of keeping jobs and getting raises, (3) investor behavior pulling money in and out of categories to chase returns and seemingly persistently get the timing wrong as a group, meaning there exist others on the opposite sides of those trades benefiting, (4) behavioral biases, which persist in many fields even though they are known and out in the open, (5) the fact that many money managers don't in fact believe in or invest based on academic and other ideas in practice, among other things] Nobody tilting or deviating from the market should be overconfident; many will be dead wrong and pay extra money for the privilege to be wrong. Moreover, markets change.
longinvest wrote:
lack_ey wrote:Even in a market where you don't have unique information, if there is some feature about the conditions and setup—e.g. people seem to bet too frequently on speculative small growth stocks with zero profitability, resulting in low risk-adjusted returns historically—this may translate to an actionable edge if you suspect that certain features seen in the past have some chance of continuing to some degree in the future. No guarantees, but maybe you're doing better than flipping a coin now. These things are based on return expectations, that certain stocks will do worse than other ones, even. Never mind the much less controversial positions of "stocks have higher expected returns than bonds, probably."
Demand and supply does not rule out the possibility that a segment of the market is priced, often enough, so as to deliver higher returns, if it exposes investors to higher volatility and risk. But, I do not believe in free lunches.

In other words, one can always get a chance at even higher returns by taking the risk of even lower returns. There are many ways to do that, including tilting and leverage.

I am relatively conservative. I prefer to be the tortoise than the hare, even if it's not flashy.

I do believe in mathematics, though. So, I know that when I invest in a low-cost total-market index fund, I am guaranteed to reap market returns, whatever they are. Not only that, but I'll also beat the active investors of that market in the aggregate, after fees. (Sharpe's theorem).
Let's forget about securities within asset classes. In a relatively reasonable market, across stocks and bonds, what is the equilibrium level of expected return for each asset class? Hopefully the riskier assets like stocks return more... right? It need not follow the capital markets line or any such abstraction, but people should demand higher return for higher risk in some reasonable set of financial assets.

This seems to agree with "Demand and supply does not rule out the possibility that a segment of the market is priced, often enough, so as to deliver higher returns, if it exposes investors to higher volatility and risk. But, I do not believe in free lunches."

Even without any free lunches, this states that at times the market will price stocks for higher expected returns (because of the higher risks) than bonds. So even if explicitly your framework doesn't necessarily by some definitions rely on return predictions, it is based on principles and understandings (that there are no free lunches for risk/return) that implicitly make use of them.

I mean, it would seem logically contradictory to believe (1) there are no free lunches, (2) bonds have lower risks and volatility than stocks, and (3) bonds have greater or equal expected returns as stocks. What kind of lunch would that be? If bonds weren't expected to return less and were less risky, the market would take the free lunch and act to pile into more bonds, bringing yields back down. As such, because you stated the first two, the third point must be false. And if that's false, that means that you think stocks have greater expected returns than bonds. Either that, or you have a definition of free lunches that I don't understand.
BahamaMan
Posts: 896
Joined: Wed Oct 01, 2014 5:52 pm

Re: The Futility of Predicting Future Returns

Post by BahamaMan »

lack_ey wrote: I mean, it would seem logically contradictory to believe (1) there are no free lunches, (2) bonds have lower risks and volatility than stocks, and (3) bonds have greater or equal expected returns as stocks. What kind of lunch would that be? If bonds weren't expected to return less and were less risky, the market would take the free lunch and act to pile into more bonds, bringing yields back down. As such, because you stated the first two, the third point must be false. And if that's false, that means that you think stocks have greater expected returns than bonds. Either that, or you have a definition of free lunches that I don't understand.
And of course no one said this. That is why you don't understand.
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

BahamaMan wrote:
lack_ey wrote: I mean, it would seem logically contradictory to believe (1) there are no free lunches, (2) bonds have lower risks and volatility than stocks, and (3) bonds have greater or equal expected returns as stocks. What kind of lunch would that be? If bonds weren't expected to return less and were less risky, the market would take the free lunch and act to pile into more bonds, bringing yields back down. As such, because you stated the first two, the third point must be false. And if that's false, that means that you think stocks have greater expected returns than bonds. Either that, or you have a definition of free lunches that I don't understand.
And of course no one said this. That is why you don't understand.
On the contrary,
longinvest wrote:Demand and supply does not rule out the possibility that a segment of the market is priced, often enough, so as to deliver higher returns, if it exposes investors to higher volatility and risk. But, I do not believe in free lunches.
longinvest wrote:bonds have lower price volatility than stocks (they can't gain value as fast as stocks, nor lose value as fast as stocks)
[...]
Mathematics tell us about the limits on the volatility of an aggregate bond fund. What do we know about the human psychology that determines the price of a stock? How can we predict the future success and profits of a company?
[...]
Stocks are effectively risky, but their volatility can go in the right direction (e.g. up!). Also, stocks do not live in a vacuum; they are accessible to the same investors that invest in bonds. If few people invested in profitable companies, the price of their stocks would fall down and the audacious investors that bought them would reap the rewards (high dividends). Other investors would notice, get jealous, and start buying these stocks, and thus push their price up. That's how stocks end up being volatile.
As far as I can tell "I do not believe in free lunches" in the context of risk satisfies (1). Then the second set of quotes satisfies (2). My own statement of (3) is made up to be logically inconsistent with (1) or (2) unless you can show me how that isn't right. If we assume (1) and (2) to be true, this means that (3) is false. And if (3) is false, this means that bonds do not have "greater or equal expected returns as stocks"—that is, they have lower expected returns as stocks.

The implication is that anyone who believes (1) and (2) should believe that stocks have greater expected returns than bonds, or alternatively something else, such as that it is not a free lunch for there to be two large asset classes, one of which is riskier than the other but doesn't offer extra returns.

Maybe I'm reaaaally stupid and missed something. Help?
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

lack_ey wrote:
longinvest wrote: I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
Okay, that is fair enough. I dropped a couple words here and there too, so I can't say anything. I just wanted to make sure I got what you were after.
You still don't provide an actionable alternative, though.
lack_ey wrote:
I mean, it would seem logically contradictory to believe (1) there are no free lunches, (2) bonds have lower risks and volatility than stocks, and (3) bonds have greater or equal expected returns as stocks.
1- For one thing, I have not used the statistical term "expected return" in my OP, because I am not a statistician and I know that its semantics is not what normal people think. Worse, an "expected return" should probably always be accompanied with an range (and possibly a confidence level, like polls). Right?

2- Where did I write (3)? I don't remember writing that.
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

lack_ey,

I will not reply to your comments related to a hypothetical (unactionable) example that assumes being privy to some priviledged information. In Canada, it is illegal to trade on the markets taking advantage of such information. I am relatively confident that it is also illegal to do so in the U.S.
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

longinvest wrote:
lack_ey wrote:
longinvest wrote: I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
Okay, that is fair enough. I dropped a couple words here and there too, so I can't say anything. I just wanted to make sure I got what you were after.
You still don't provide an actionable alternative, though.
I noted earlier that one may want to use CDs when they provide attractive yields relative to the expected value and risks of bonds based on a statistical advantage. No guarantees, but I think you have a good shot, over 50%, of doing better at times.

Another example was more recent, to potentially avoid speculative small growth stocks with (what I believe might be) low expected returns, particularly for the risk. Fortunately, many of these are new IPOs that aren't even included in most indexes, not meeting the inclusion screens. And the market cap is generally low enough not to make that much a difference anyway even if you cap weighted everything and had a fund that covered 100% of the space.

In addition, let's say I felt that I was 10 years out from retirement and had 90% of the value I wanted built up, and bond expected returns are in the 3% real range (let's just say that 10-year TIPS yield 3%) and stocks seemingly overvalued with a dividend yield of 1% and CAPE of 50 (and no more buybacks than today and no real evidence of superior long-term growth). I would decide what I think the expected returns of stocks and bonds are and probably go to something like 20/80 stocks/bonds. No need to take the extra risk and furthermore no desire to, with the risk/return of stocks being relatively unattractive (is what I'm betting on, no guarantees).* If stocks were instead yielding 3% with a CAPE of 10 in that example, I would consider something more like 50/50 and a wide variety of options to be reasonable.

*Here I would say that the market probably still thinks stocks will return more than bonds, but that they seem not to be demanding much extra return for the risk. In other words, a reasonable market may have different preferences for risks and price in different things at different times.
longinvest wrote:
lack_ey wrote:
I mean, it would seem logically contradictory to believe (1) there are no free lunches, (2) bonds have lower risks and volatility than stocks, and (3) bonds have greater or equal expected returns as stocks.
1- For one thing, I have not used the statistical term "expected return" in my OP, because I am not a statistician and I know that its semantics is not what normal people think. Worse, an "expected return" should probably always be accompanied with an range (and possibly a confidence level, like polls). Right?
I won't say "always" but confidence intervals are nice and certainly helpful, as are other depictions of the uncertainty. You are right that a lot of people, particularly without heavier math backgrounds, may be unnecessarily confused.

Then there also needs to be the understanding from the reader that even if a model predicts a certain range of possibilities, the model could well be wrong in various ways. The best models are the kinds that are tested over and over again and refined. If you say that results should fall withing X and Y at a rate of 75%, in the long run the actual results better fall in the predicted range around 75% of the time—not 50% of the time, and not 90% either.
longinvest wrote:2- Where did I write (3)? I don't remember writing that.
Please see the above post. I only mean that you wrote (1) and (2). I constructed (3) to logically clash with (1) and (2). This is a vain attempt as part of the "proof" of some result in the logical/mathematic sense. If in general (3) cannot be true if (1) and (2) are true, then if we believe (1) and (2) to be true, we logically must believe (3) to be false.
BahamaMan
Posts: 896
Joined: Wed Oct 01, 2014 5:52 pm

Re: The Futility of Predicting Future Returns

Post by BahamaMan »

lack_ey wrote:
longinvest wrote:2- Where did I write (3)? I don't remember writing that.
Please see the above post. I only mean that you wrote (1) and (2). I constructed (3) to logically clash with (1) and (2). This is a vain attempt as part of the "proof" of some result in the logical/mathematic sense. If in general (3) cannot be true if (1) and (2) are true, then if we believe (1) and (2) to be true, we logically must believe (3) to be false.
Yup, and that's why I said that no one said that. Constructing things no one said is exactly why you are confused.
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

Let me try one last time. Either I'm having an RBD or there's a failure of logic somewhere.

These things were said by longinvest, paraphrased:
(1) no free lunch
(2) stocks have higher volatility than bonds
(X) I am not making return assumptions, not considering future asset class returns

Let us assume that (1) and (2) are true. Then consider the following statement:
(3) bonds have greater or equal expected returns than stocks

I claim that if (1) and (2) are true, (3) cannot also be true. That is, it is false that bonds have greater or equal expected returns than stocks. This is the same as saying that stocks have greater expected returns than bonds.*

Therefore I claim that anyone who believes (1) and (2) thinks that stocks have greater expected returns than bonds. This belief is in contradiction with (X), which states that there are no return assumptions. Embedded in (1) and (2) are a belief that stocks are priced reasonably compared to bonds, which implies that stocks generally have higher risk and return. So statements (1), (2), and (X) are in contradiction with each other, if you follow and verify all the steps.


*This follows from the idea that a relatively non-braindead market would not allow bonds to both return more and be less risky than stocks. Who would own stocks, then, outside of speculators? Investors would pour into those bonds until the yields went down, to the point of eliminating this hypothetical free lunch.
User avatar
Taylor Larimore
Advisory Board
Posts: 30457
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Boglehead Contests

Post by Taylor Larimore »

Bogleheads:

Many years ago I started the Boglehead Contests to demonstrate how difficult it is to forecast stock market returns.

I pay no attention to stock market forecasts. They are meaningless to a buy and hold investor -- and usually wrong. We get what the market gives.

Best wishes.
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

lack_ey wrote:
longinvest wrote:
lack_ey wrote:
longinvest wrote: I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
Okay, that is fair enough. I dropped a couple words here and there too, so I can't say anything. I just wanted to make sure I got what you were after.
You still don't provide an actionable alternative, though.
I noted earlier that one may want to use CDs when they provide attractive yields relative to the expected value and risks of bonds based on a statistical advantage. No guarantees, but I think you have a good shot, over 50%, of doing better at times.
OK. Help me understand.

You are saying that I could buy a 5-year, CD today, and have a good shot, over 50%, of beating the return of Vanguard's Total Bond Market Index fund (VBTLX) over the next 5 years.

I'd like to see your detailed calculations. I want you to show me your spreadsheet with the numerical information of every bond in BND (based on the latest financial statement), and your simulation framework for all (or a significant enough sample) of the reasonably possible yield curves (for reinvestment yields) in the next five years. You should also let me know how you estimated the possible bond upgrades and downgrades that will happen. Don't forget to also tell me how you modeled the various possible changes in the weight of various types of bonds in the market during the period.

Otherwise, you're just telling me that you like doing "mild speculation" using CDs in the fixed income part of your portfolio, based on some technical indicators. Great! But, this does not provide an attractive alternative to simply put the money in VBTLX and rebalancing with stocks once a year.

Let me go even one step further, and assume that your CDs will beat (100%) VBTLX. By how much? 1%? Do you think that this once-in-a-blue-moon 1% will make or break my retirement?

And, there are other considerations, still. Chasing CDs and managing multiple accounts across various institutions would add complexity to my portfolio, and make it much more difficult to manage for my wife, if something ever happened to me.

So far, the OP's good enough portfolio wins.
lack_ey wrote:Another example was more recent, to potentially avoid speculative small growth stocks with (what I believe might be) low expected returns, particularly for the risk. Fortunately, many of these are new IPOs that aren't even included in most indexes, not meeting the inclusion screens. And the market cap is generally low enough not to make that much a difference anyway even if you cap weighted everything and had a fund that covered 100% of the space.
How is this actionable, as an alternative to my OP's suggestion, to set an asset allocation?
lack_ey wrote:In addition, let's say I felt that I was 10 years out from retirement and had 90% of the value I wanted built up, and bond expected returns are in the 3% real range (let's just say that 10-year TIPS yield 3%) and stocks seemingly overvalued with a dividend yield of 1% and CAPE of 50 (and no more buybacks than today and no real evidence of superior long-term growth). I would decide what I think the expected returns of stocks and bonds are and probably go to something like 20/80 stocks/bonds. No need to take the extra risk and furthermore no desire to, with the risk/return of stocks being relatively unattractive (is what I'm betting on, no guarantees).* If stocks were instead yielding 3% with a CAPE of 10 in that example, I would consider something more like 50/50 and a wide variety of options to be reasonable.

*Here I would say that the market probably still thinks stocks will return more than bonds, but that they seem not to be demanding much extra return for the risk. In other words, a reasonable market may have different preferences for risks and price in different things at different times.
Your predictions are misleading, because of their implicit tight ranges. (The proof is in my OP). How could I trust that VBTLX will be in the 3% range; that's too tight (implicit what? +/- 0.1%? +/- 0.5%? if it was bigger, you would have spelled it out, right?).

I want something way more explicit. Something like: a cummulative total return of 35% +/- 60% (which makes the 10-year error quite explicit), or the equivalent: a compound annual return of 3% +/- 4%.

Interesting, isn't it, that a small +/- 4% error translates into a +/- 60% error (a 120% range) over 10 years? :wink:
lack_ey wrote:
longinvest wrote:2- Where did I write (3)? I don't remember writing that.
Please see the above post. I only mean that you wrote (1) and (2). I constructed (3) to logically clash with (1) and (2). This is a vain attempt as part of the "proof" of some result in the logical/mathematic sense. If in general (3) cannot be true if (1) and (2) are true, then if we believe (1) and (2) to be true, we logically must believe (3) to be false.
This doesn't add anything new to this thread. I've have already written, in an earlier post:
viewtopic.php?f=10&t=170335#p2568295
longinvest wrote:So, can an investor rationally hope that stocks will have a chance at getting higher returns than bonds? Yes, he can hope for it. If it didn't happen often enough, people would shun stocks and their prices would get lower until they would eventually offer better forward returns than bonds. But, the investor does not know if he will get such returns.
In other words, stocks are sometimes cheap, but sometimes expensive. It is unknowable, at any moment, how they are (cheap or expensive) with sufficient accuracy to predict whether stocks or bonds will outperform in a specific future time period. Also, if stocks were not cheaper than bonds often enough, investors would shun them (go into bonds) and their price would get lower. Investors want higher returns as compensation for higher risk. Basic law of demand and supply. But "risk" means that investors don't always get their wishes...

The thing is that the law of demand and supply is quite annoying, for making future return predictions. If it was possible, using simple techniques such as P/E or DDM, to accurately forecast future stock returns, then there would not be risk, anymore, for investing in stocks. So, no sane investor would invest in bonds which return less than stocks. It cannot be otherwise, if enough investors have any brains.

So, either (1) you have found a sophisticated and precise prediction method, that nobody knows about, but I am sure that you'll be glad to share with me, or (2) other investors know about these valuation techniques and are making the same predictions.

You'll have to forgive me; while I think that you are a very nice and brilliant person, I can't bring myself to believe in (1). So, I'm left with (2), which means that, either bonds and stocks are fairly priced, if stocks are not riskier than bonds, or stocks have a significant chance of being more expensive or cheaper than your valuation metric is predicting. So, the prediction is not actionable.
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: Boglehead Contests

Post by longinvest »

Taylor Larimore wrote:Bogleheads:

Many years ago I started the Boglehead Contests to demonstrate how difficult it is to forecast stock market returns.

I pay no attention to stock market forecasts. They are meaningless to a buy and hold investor -- and usually wrong. We get what the market gives.

Best wishes.
Taylor
Thank you, Taylor.

I admire your talent for saying things clearly and concisely.

Best regards,

longinvest
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
jginseattle
Posts: 836
Joined: Fri Jul 01, 2011 7:33 pm

Re: The Futility of Predicting Future Returns

Post by jginseattle »

"Job one for any investor is to estimate asset class returns. For the first time, Antti Ilmanen has assembled into one volume all of the tools necessary for this task: for the working money manager, a unique treasure trove of analytical techniques and empirical evidence; for the academic, a comprehensive guide to the relevant academic literature; and for the consultant, a blinding light with which to illuminate performance. Expected Returns is destined to occupy the front shelves of investment professionals around the world."

William J. Bernstein, author of The Intelligent Asset Allocator, The Birth of Plenty, and A Splendid Exchange, and co-principal of Efficient Frontier Advisors.

(From the book jacket).
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

jginseattle wrote:"Job one for any investor is to estimate asset class returns. For the first time, Antti Ilmanen has assembled into one volume all of the tools necessary for this task: for the working money manager, a unique treasure trove of analytical techniques and empirical evidence; for the academic, a comprehensive guide to the relevant academic literature; and for the consultant, a blinding light with which to illuminate performance. Expected Returns is destined to occupy the front shelves of investment professionals around the world."

William J. Bernstein, author of The Intelligent Asset Allocator, The Birth of Plenty, and A Splendid Exchange, and co-principal of Efficient Frontier Advisors.

(From the book jacket).
Great! So, I can count on thousands and thousands of institutional money investors to build equilibrium in market pricing and arbitrage any mispricing, right?

Why should anyone or any fund company that has discovered a method to beat the market be benevolent and share the profits with me?
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

longinvest wrote:
lack_ey wrote:
longinvest wrote:
lack_ey wrote:
longinvest wrote: I have yet to see anybody, in this thread, provide an actionable alternative, like a better way to construct a portfolio using return predictions that is guaranteed to outperform the proposed good enough portfolio built without return predictions.
Okay, that is fair enough. I dropped a couple words here and there too, so I can't say anything. I just wanted to make sure I got what you were after.
You still don't provide an actionable alternative, though.
I noted earlier that one may want to use CDs when they provide attractive yields relative to the expected value and risks of bonds based on a statistical advantage. No guarantees, but I think you have a good shot, over 50%, of doing better at times.
OK. Help me understand.

You are saying that I could buy a 5-year, CD today, and have a good shot, over 50%, of beating the return of Vanguard's Total Bond Market Index fund (VBTLX) over the next 5 years.

I'd like to see your detailed calculations. I want you to show me your spreadsheet with the numerical information of every bond in BND (based on the latest financial statement), and your simulation framework for all (or a significant enough sample) of the reasonably possible yield curves (for reinvestment yields) in the next five years. You should also let me know how you estimated the possible bond upgrades and downgrades that will happen. Don't forget to also tell me how you modeled the various possible changes in the weight of various types of bonds in the market during the period.

Otherwise, you're just telling me that you like doing "mild speculation" using CDs in the fixed income part of your portfolio, based on some technical indicators. Great! But, this does not provide an attractive alternative to simply put the money in VBTLX and rebalancing with stocks once a year.

Let me go even one step further, and assume that your CDs will beat (100%) VBTLX. By how much? 1%? Do you think that this once-in-a-blue-moon 1% will make or break my retirement?

And, there are other considerations, still. Chasing CDs and managing multiple accounts across various institutions would add complexity to my portfolio, and make it much more difficult to manage for my wife, if something ever happened to me.

So far, the OP's good enough portfolio wins.
I haven't run the numbers because I don't have enough money in fixed income now for it to be worth it (with a relocation looming, current savings).

You're also creating a straw man with respect to running all the calculations and scenarios for every credit upgrade and downgrade. I never implied anyone had to do that, just that some implicit or explicit estimates of expected returns can be used to make beneficial decisions.

As I said earlier, we know what the market prices a 5-year bond backed by the government at. The rate you can find on a CD will beat that if you buy and hold both to maturity.

It stands to reason that other bonds on the market are (approximately, in the least) priced appropriately relative to 5-year Treasuries; those with higher yields will come with higher risks. Total bond market comes with a little more credit risk than a CD and also a slightly higher starting duration and also increasingly higher duration over the holding period. The duration is about 5.5 years to start compared to a bit under 5 years for the CD. On the last day of the CD before maturity, the CD has a duration of 0 and the fund still has a duration around that 5.5 years.

So I am fairly confident the CD has superior risk/return properties. Looking at expected return for a bond fund and not just a 5-year Treasury, you'd check SEC yield, maybe expected rolldown yield if rates stay the same, etc. It's quite possible now that total bond returns more, maybe even over 50%. But that's with more risk. Look at an intermediate-term Treasury bond fund, which doesn't have the credit risk, and I'd give the CD over 50% to earn more, and that's taking more term risk in the fund.

On average with relatively good CD deals you might find an edge of expected return under 1% just for the fixed income side, and that's ignoring interactions with stocks. In practice the overall benefit may be under 0.25% from a whole-portfolio perspective on average over the years, even with relatively heavy fixed income allocations. That's not going to make or break a retirement. Yes, it's extra complexity and a hassle, but you asked for something actionable.

(Though really, you could use a brokered CD if you're really docking this for the hassle factor; you lose the early withdrawal provision that might be available elsewhere, and lose that liquidity and option. But you still have returns nontrivially above a 5-year Treasury without much additional risk.)
longinvest wrote:
lack_ey wrote:Another example was more recent, to potentially avoid speculative small growth stocks with (what I believe might be) low expected returns, particularly for the risk. Fortunately, many of these are new IPOs that aren't even included in most indexes, not meeting the inclusion screens. And the market cap is generally low enough not to make that much a difference anyway even if you cap weighted everything and had a fund that covered 100% of the space.
How is this actionable, as an alternative to my OP's suggestion, to set an asset allocation?
It's usually not, but if you look under the hood of a given total market fund and find these investments, it may possibly be worth using some other fund instead, if you estimate the drag here to be nontrivial (it's probably largely trivial compared to a total market fund, though potentially not trivial for specifically a small-cap growth fund). Again, it's possible that the drag is a mirage and doesn't show up. You never know.
longinvest wrote:
lack_ey wrote:In addition, let's say I felt that I was 10 years out from retirement and had 90% of the value I wanted built up, and bond expected returns are in the 3% real range (let's just say that 10-year TIPS yield 3%) and stocks seemingly overvalued with a dividend yield of 1% and CAPE of 50 (and no more buybacks than today and no real evidence of superior long-term growth). I would decide what I think the expected returns of stocks and bonds are and probably go to something like 20/80 stocks/bonds. No need to take the extra risk and furthermore no desire to, with the risk/return of stocks being relatively unattractive (is what I'm betting on, no guarantees).* If stocks were instead yielding 3% with a CAPE of 10 in that example, I would consider something more like 50/50 and a wide variety of options to be reasonable.

*Here I would say that the market probably still thinks stocks will return more than bonds, but that they seem not to be demanding much extra return for the risk. In other words, a reasonable market may have different preferences for risks and price in different things at different times.
Your predictions are misleading, because of their implicit tight ranges. (The proof is in my OP). How could I trust that VBTLX will be in the 3% range; that's too tight (implicit what? +/- 0.1%? +/- 0.5%? if it was bigger, you would have spelled it out, right?).

I want something way more explicit. Something like: a cummulative total return of 35% +/- 60% (which makes the 10-year error quite explicit), or the equivalent: a compound annual return of 3% +/- 4%.

Interesting, isn't it, that a small +/- 4% error translates into a +/- 60% error (a 120% range) over 10 years? :wink:
We now have pages of discussion and graphs showing what kind of results are likely. I didn't think it necessary to repeat them all. The expected value by definition is just the probability-weighted average, a single value. I left it up to the reader to infer the ranges of possible outcomes based on all the other info.

If it makes you feel better, maybe I replace the "bonds" there with a TIPS ladder for the example?

Again, I never said that explicit return assumptions were necessary. If I had to guess on the spot, I would say an expected value somewhere around 3% real annualized over 10 years, with a 75% chance between 1.5% and 4.5% annualized real (for the 20/80). Obviously I just made that up.
longinvest wrote:
lack_ey wrote:
longinvest wrote:2- Where did I write (3)? I don't remember writing that.
Please see the above post. I only mean that you wrote (1) and (2). I constructed (3) to logically clash with (1) and (2). This is a vain attempt as part of the "proof" of some result in the logical/mathematic sense. If in general (3) cannot be true if (1) and (2) are true, then if we believe (1) and (2) to be true, we logically must believe (3) to be false.
This doesn't add anything new to this thread. I've have already written, in an earlier post:
viewtopic.php?f=10&t=170335#p2568295
longinvest wrote:So, can an investor rationally hope that stocks will have a chance at getting higher returns than bonds? Yes, he can hope for it. If it didn't happen often enough, people would shun stocks and their prices would get lower until they would eventually offer better forward returns than bonds. But, the investor does not know if he will get such returns.
Then it is just another example that you seem to be avoiding the notion that return assumptions are baked into ideas about asset class behavior. In any case, you need an understanding of what asset classes do to have any idea about what reasonable portfolio constructions might look like. This just makes the OP look confusing (to me, and several others who have posted), then.

longinvest wrote:The thing is that the law of demand and supply is quite annoying, for making future return predictions. If it was possible, using simple techniques such as P/E or DDM, to accurately forecast future stock returns, then there would not be risk, anymore, for investing in stocks. So, no sane investor would invest in bonds which return less than stocks. It cannot be otherwise, if enough investors have any brains.

So, either (1) you have found a sophisticated prediction method, that nobody knows about, but I am sure that you'll be glad to share with me, or (2) other investors know about these valuation techniques and are making the same predictions.

You'll have to forgive me; while I think that you are a very nice and brilliant person, I can't bring myself to believe in (1). So, I'm left with (2), which means that, either bonds and stocks are fairly priced, if stocks are not riskier than bonds, or stocks have a possibility of being more expensive (or cheaper) than your valuation metric is predicting.
Nobody's disagreeing that actual returns are unknown and that there is a lot of uncertainty. Investing is about playing the odds responsibly to try to achieve desired outcomes, using explicit calculations or not. Whether or analyze it this way, whichever allocation you choose has a certain chance of doing certain things, including perhaps not sustaining the kind of buying power or growth you'd like to see. I just suggest that there is some information out there about the current and past states of the market that can be acted on to improve your odds in some cases.

This doesn't necessarily involve any market-beating prediction methods. Sometimes it might, possibly, though fewer people agree with that. In the more general scenario you can look at relative asset class pricing to gain some limited information about how the distributions may have shifted compared to what has historically been seen. Most of the times (especially years away from needing the money) it's not really going to result in much meaningful action to be taken if you have something like a three-fund portfolio, and a static allocation may be better or just about the same in the long run. It's more useful for vague planning purposes than asset allocation decisions.
User avatar
Maynard F. Speer
Posts: 2139
Joined: Wed Mar 18, 2015 10:31 am

Re: The Futility of Predicting Future Returns

Post by Maynard F. Speer »

Here are two charts I think demonstrate the usefulness of valuation metrics

From Hussman:
Image

And with global markets:
Image

All valuation does here is give you a contextualised historical market return .. It adds a Y-axis .. As history's demonstrated over and over again: Being in the highest growth stock, sector or region has very little predictive power over long-term returns - what you pay for something has a fairly strong relationship

--
"Actionable" ideas:

Meb Faber

1) At a minimum, allocate your portfolio globally reflecting the global market cap weightings. In the U.S., that means allocating 50% of your portfolio abroad.

2) To avoid market cap concentration risk, consider allocating along the weightings of global GDP. This would mean closer to 60-80% in foreign stocks.

3) Similarly, ponder a value approach to your equity allocation. Consider over-weighting the cheapest countries and avoiding the most expensive ones. Currently, this would mean a low, or zero, allocation to U.S. stocks.

- From Global Value
"Economics is a method rather than a doctrine, an apparatus of the mind, a technique of thinking, which helps its possessor to draw correct conclusions." - John Maynard Keynes
User avatar
Wildebeest
Posts: 1204
Joined: Fri Dec 27, 2013 2:36 pm

Re: The Futility of Predicting Future Returns

Post by Wildebeest »

When we came into money, we looked into real estate, buying another business etc. The equity market looked great and with little work and little cost, we would get close to a 10 % return on investment. We decided to invest in 2007 in the stockmarket and actually it looked so profitable, I was leveraged as well before the bottom fell out.

Currently I am convinced it will be considerably less for the foreseeable future. I am not watching the stock market, in 2008 and 2009 I actually stopped looking at our holdings, because it was too depressing. Currently it is wonderful but I am still shell shocked. I have learned my lesson. There is no rhyme or reason.

I have no idea what else to invest in, which requires so little effort and provides such a great return, be it 3 %, 6 % or 10 %. Per our investment philosophy statement: If the 10 year T bill rate goes over 5 % we will transfer into bonds. Will it be smart decision: I doubt it, but it was made with predictions about future returns in 2007.

We are set and I doubt we will change our investment philosophy statement in the near or far future what ever predictions abound.

I still enjoy reading about predictions and I am grateful for the predictions in 2007 of future returns. Now the predictions are futile since they are not actionable, but they were very meaningful in 2007.
The Golden Rule: One should treat others as one would like others to treat oneself.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Re: The Futility of Predicting Future Returns

Post by grayfox »

Suppose that I am 65 years old and have $1,000,000 portfolio invested 50/50 in TSM/TBM. I will fund my 30-year retirement by withdrawing from the portfolio. I am willing to spend down the portfolio. How much should I withdraw each year?

Of course, everyone is familiar with the 4% + inflation withdrawal. So I would withdraw $40,000 the first year and adjust for inflation each year. Now a lot of people don't like this method because it ignores the actual returns you get. Bad returns and you run out of money.

:idea: Some experts have suggested using the Excel PMT() function to calculate the withdrawal amount each each. PMT(), one of Excel's financial functions, calculates the constant period payment required to pay off a loan or investment over some period. The inputs to payment are rate=rate of return, nper=number of periods, pv=present value, fv=future value, and type. Each year I will run PMT() to calculate how much I should withdraw that year.

E.g. If the rate of return is 2%, for the first year I should withdraw PMT(0.02, 40, -1000000, 0, 0) = 36556

Is anyone familiar with this method?

:?: My question is: What rate of return should I use for my 50/50 TSM/TBM in the PMT() function? I'm not sure if I should use the E(return), the median, 2.5%-tile worst case or what. But it seems that I have to come up with some prediction of future return to use the PMT() method.
IlliniDave
Posts: 2369
Joined: Fri May 17, 2013 7:09 am

Re: The Futility of Predicting Future Returns

Post by IlliniDave »

lack_ey wrote:Let me try one last time. Either I'm having an RBD...
Nah, you're doing fine. I understand exactly what you are saying. Siamond observed earlier that this topic can begin to take on a vibe similar to religious belief. Essentially it all comes down to valuations and risk and some people just don't believe that valuations matter.

Whether or not something is actionable can be in the eye of the beholder. If a weather forecast says rain is possible a person who already plans on staying indoors all day no matter what could state, "It's not actionable, forecasting today's weather is therefore futile." But for someone who plans on spending time outdoors that day, it is actionable, even if it's just the small action of carrying an umbrella.

I think that's sort of where we are at in this discussion.

From my own perspective all I can say is this. Other than some small changes in how I evaluate my "plan", things related to equity outlooks as they stand today haven't prompted me to take any action regarding investment portfolio decisions. But we're creeping towards a point I might. There can be a correlation between low forward-looking return outlooks and unusually high valuations. Should the higher valuation/lower outlook trend continue, at some point my inherent risk tolerance could be violated in which case I'll back off equities by 10%-20%. So to me monitoring the valuation/return outlook space is actionable (and that's all in my "policy"). Some people will accuse me of market-timing and trying to "beat the market" and I don't care if they do. My real intent is to balance risk with my risk tolerance, neither of which are static precisely-known quantities.

For an investor whose strategy is to choose an AA a priori and stick to it no matter what, the valuation/return outlook space contains nothing actionable.

Both sides can construct rationale to support their position. Both approaches are reasonable. Different temperaments will gravitate towards one or the other.
Don't do something. Just stand there!
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

grayfox wrote:Suppose that I am 65 years old and have $1,000,000 portfolio invested 50/50 in TSM/TBM. I will fund my 30-year retirement by withdrawing from the portfolio. I am willing to spend down the portfolio. How much should I withdraw each year?

Of course, everyone is familiar with the 4% + inflation withdrawal. So I would withdraw $40,000 the first year and adjust for inflation each year. Now a lot of people don't like this method because it ignores the actual returns you get. Bad returns and you run out of money.

:idea: Some experts have suggested using the Excel PMT() function to calculate the withdrawal amount each each. PMT(), one of Excel's financial functions, calculates the constant period payment required to pay off a loan or investment over some period. The inputs to payment are rate=rate of return, nper=number of periods, pv=present value, fv=future value, and type. Each year I will run PMT() to calculate how much I should withdraw that year.

E.g. If the rate of return is 2%, for the first year I should withdraw PMT(0.02, 40, -1000000, 0, 0) = 36556

Is anyone familiar with this method?

:?: My question is: What rate of return should I use for my 50/50 TSM/TBM in the PMT() function? I'm not sure if I should use the E(return), the median, 2.5%-tile worst case or what. But it seems that I have to come up with some prediction of future return to use the PMT() method.
Grayfox,

I'll try to provide a partial answer to your question, but you should take any follow-up to another thread, as the subject of retirement planning (as I will explain) is orthogonal to the issue of not needing future return predictions to select an asset allocation and construct a good enough portfolio (the topic of this thread).

But, first, let me clarify things that I did write and others that I did not write in my first post.

----

In my post, I wrote that one does not need to use future return predictions to pick an asset allocation and invest into a Three-Fund Portfolio; that the resulting portfolio would be good enough.

I did not write that stocks and bonds will have the same future returns.

I did not write that one cannot know the exact nominal internal rate of return of a Treasury bond (assuming coupons are not reinvested and no default), bought at issue and held to maturity, simply by looking at its coupon rate. This does not require making a prediction.

I did not write that one cannot know the exact real internal rate of return of a TIPS (assuming coupons are not reinvested and no default), bought at issue and held to maturity, simply by looking at its coupon rate. This does not require making a prediction.

I did prove that predicting the precise future return of a total-market bond fund is impossible.

I did prove that predicting with high or moderate precision the future return of a total-market stock fund is impossible.

I did not prove that making any future return prediction, even if imprecise, is impossible.

I did write that an imprecise future return prediction should be accompanied with a range, otherwise it is misleading.

I did not write that riskier assets should not command a lower price most of the time (but not always). Actually, I've been quite explicit in follow-up posts: we can derive from the law of demand and supply that it would be illogical for stocks not to be priced often enough so that their future returns are higher than those of [short-term] bonds, because if it wasn't so, no sane investor would invest in stocks. But, this can only be possible if stocks are, at other times, priced so that their future returns are lower, otherwise stocks wouldn't be riskier than short-term bonds (which is a reasonable hypothesis to make, due to the nature of stocks and bonds).

Let me add that long-term nominal bonds do have a risk of trailing future inflation. The law of supply and demand applies to them, too, as it applies to all assets.

----

You are asking a question about planning retirement. (This is an entirely different topic.)

Of course, some people use simplistic and unrealistic retirement planning models, and as a consequence derive useless conclusions.

An example of a simplistic and unrealistic retirement planning model is to assume that:
  1. Social Security does not exist.
  2. Inflation-indexed Social Security payments cannot be increased by delaying it to 70 years old.
  3. One cannot cover the period between retirement and the start of Social Security payments using fixed-income products, cash, or a term-certain single premium immediate annuity.
  4. Life single premium immediate annuities (SPIA), indexed to inflation or not, do not exist.
  5. It is impossible to calculate the exact cost of a non-rolling TIPS ladder that delivers a fixed, inflation-adjusted amount of money (coupons + principal) every year for up to 30 years long. (#Cruncher has not developed a tool to do it).
  6. The only possible retirement model is to make a fixed inflation-adjusted withdrawal every year from a portfolio. The starting ratio of withdrawal to portfolio balance is called a Safe Withdrawal Rate (SWR).
  7. Other portfolio withdrawal methods do not exist.
  8. All retirements are exactly 30-years long.
  9. One cannot combine delaying Social Security, pensions, life SPIAs, and a portfolio using a flexible withdrawal method to plan retirement.
Now, if you make all of these unrealistic assumptions, of course it becomes critical to make precise future return predictions, otherwise you'll risk eating Alpo* in retirement, even if you are a multi-millionaire. So much anxiety!

* Alpo is a brand of pet food, used for illustrative purpose.

Instead, I invite you to revise the list of assumptions and eliminate all of them. You'll realize that combining delayed Social Security, pensions, life SPIAs, a flexible withdrawal method, and a good enough Three-Fund Portfolio works.

There is simply no reason to develop high anxiety about retirement by seeking to resolve the impossible: predicting exact future returns and your exact life span.


Most of the flexible portfolio withdrawal methods only require an imprecise future return prediction to be set up. That's the case, for example, of VPW. So, again, there is not contradiction with anything I have written in this thread.

I hope that this helps you. If you want to further discuss the subject, I invite you to start a new thread.

longinvest
Last edited by longinvest on Wed Jul 29, 2015 9:21 am, edited 1 time in total.
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
User avatar
SimpleGift
Posts: 3869
Joined: Tue Feb 08, 2011 3:45 pm
Location: Central Oregon

Re: The Futility of Predicting Future Returns

Post by SimpleGift »

IlliniDave wrote:Whether or not something is actionable can be in the eye of the beholder. If a weather forecast says rain is possible a person who already plans on staying indoors all day no matter what could state, "It's not actionable, forecasting today's weather is therefore futile." But for someone who plans on spending time outdoors that day, it is actionable, even if it's just the small action of carrying an umbrella.

I think that's sort of where we are at in this discussion.
Beautiful analogy for this rather long and fruitless discussion! The "stay at home" folks seem to have an entirely different world view from the "go outdoors" folks, with no interest in expanding their cosmology to include a thoughtful discussion of short and long-range weather forecasting.
IlliniDave
Posts: 2369
Joined: Fri May 17, 2013 7:09 am

Re: The Futility of Predicting Future Returns

Post by IlliniDave »

Simplegift wrote:
IlliniDave wrote:Whether or not something is actionable can be in the eye of the beholder. If a weather forecast says rain is possible a person who already plans on staying indoors all day no matter what could state, "It's not actionable, forecasting today's weather is therefore futile." But for someone who plans on spending time outdoors that day, it is actionable, even if it's just the small action of carrying an umbrella.

I think that's sort of where we are at in this discussion.
Beautiful analogy for this rather long and fruitless discussion! The "stay at home" folks seem to have an entirely different world view from the "go outdoors" folks, with no interest in expanding their cosmology to include a thoughtful discussion of short and long-range weather forecasting.
Just to be clear, my intent was not to promote one as superior to the other, nor imply that either viewpoint is more or less thoughtful than the other. They are just different in their outlooks.
Don't do something. Just stand there!
pkcrafter
Posts: 14823
Joined: Sun Mar 04, 2007 12:19 pm
Location: CA
Contact:

Re: The Futility of Predicting Future Returns

Post by pkcrafter »

Wow, this tread has really gotten some legs. The differing views seem to be mostly between the OP and a handful of posters who found the message incorrect enough to debate it. The great majority of Bogleheads should follow longinvest's message because it is the Boglehead way.

OK, there some Boglehead posters who has taken a long look at valuation models, expected returns, and several other considerations and they are confident enough to incorporate them into portfolio management. But let's be clear, defending and/or promoting strategies that go beyond buy and hold is not a good recommendation for most Bogleheads. Why? Because they can be successful without the complication of messing up.

Why is there any debate/discussion about valuation and timing models. If there is one method that is best, then there should be no discussion, everyone should use it. Why are there several withdrawal methods, all based on historical data and predictions? If reliable, everyone should draw the same conclusion. Why was 4% deemed the safe withdrawal rate, and then a few years ago Wade Pfau announced 1.8% was the safe rate. Why are 10 year forecasts done every year?

Did valuation artists get out of the market in 1995 and 2007? A 30 year forecast? Absurd. Why do those trying to squeeze some useful information out of historical data insist on 80 or 100 years worth of data? Because they know 20 or even 30 years is not enough. We live and invest in a series of short term conditions, and none of them will resemble long term results. Long term data will never fit present conditions, so it's only useful is some sort of vague and unreliable way.

The point I guess I'm trying to make is any timing or valuation models that result in portfolio changes isn't mainstream Boglehead philosophy. If individual investors want to do it fine, no problem. Will I be called a religious fanatic or dogmatic? I hope not, I'm just trying to rebalance the scale in this post.

Longinvest's post is good advice for new Bogleheads.

Paul
When times are good, investors tend to forget about risk and focus on opportunity. When times are bad, investors tend to forget about opportunity and focus on risk.
User avatar
SimpleGift
Posts: 3869
Joined: Tue Feb 08, 2011 3:45 pm
Location: Central Oregon

Re: The Futility of Predicting Future Returns

Post by SimpleGift »

pkcrafter wrote:The point I guess I'm trying to make is any timing or valuation models that result in portfolio changes isn't mainstream Boglehead philosophy.
Just to be clear, I haven't seen many in this thread advocating the use of "expected returns" as a timing or a valuation tool for portfolio changes. As you say, this has never been proven to be a useful practice and isn't mainstream Bogleheadism. However, expected returns can be helpful for long-range financial planning — e.g., for accumulators, to determine saving rates and work plans, and for retirees, to estimate safe withdrawal rates — which is the spirit in which Mr. Bogle and Mr. Bernstein (and many posters here, I believe) discuss expected returns.

This is the "baby" that's being chucked out with the bathwater here and that the OP seems to be ignoring.
Topic Author
longinvest
Posts: 4642
Joined: Sat Aug 11, 2012 8:44 am

Re: The Futility of Predicting Future Returns

Post by longinvest »

Simplegift wrote:
pkcrafter wrote:The point I guess I'm trying to make is any timing or valuation models that result in portfolio changes isn't mainstream Boglehead philosophy.
Just to be clear, I haven't seen many in this thread advocating the use of "expected returns" as a timing or a valuation tool for portfolio changes. As you say, this has never been proven to be a useful practice and isn't mainstream Bogleheadism. However, expected returns can be helpful for long-range financial planning — e.g., for accumulators, to determine saving rates and work plans, and for retirees, to estimate safe withdrawal rates — which is the spirit in which Mr. Bogle and Mr. Bernstein (and many posters here, I believe) discuss expected returns.

This is the "baby" that's being chucked out with the bathwater here and that the OP seems to be ignoring.
Simplegift,

I disagree with you on two points.

1- Minor point.

I have seen various posts in this thread promoting the use of future return predictions* to set an asset allocation and then make (possibly significant) allocation changes, depending on perceived market conditions. They were certainly not promoting the spirit of "stay the course"; they were at least promoting the spirit of "adjust the course to valuations".

* Not the same thing as the confusing "expected returns" term, a technical term used by statisticians which does not mean what many would intuitively think. For more details, you can look for lack_ey's explanation of the term, earlier in this thread.

2- Major point.

During my 20s, even if I had known a good prediction of the future returns of stocks and bonds, they would have been useless to me for setting a savings rate. I simply had no idea what my future salary would be over the next almost three decades. The only factor that was actionable to decide of a savings rate was my ability to live reasonably below my means and save the excess. Actually, in my early 20s, I went to university, so I had to live at my means and couldn't save money (it was still better than going into debt).

I do not think that a young investor has enough information about his future income in upcoming decades to make any useful use of future return predictions for setting a savings rate. The main deciding factor is the ability to live below one's means (but not like a pauper) and invest the rest.

When I think about it, the only use I see for using predictions, in setting the savings rate of a young investor, is to justify saving less and spending more! But, we all know that common sense should be used, instead, to balance spending and saving.
Bogleheads investment philosophy | One-ETF global balanced index portfolio | VPW
User avatar
Maynard F. Speer
Posts: 2139
Joined: Wed Mar 18, 2015 10:31 am

Re: The Futility of Predicting Future Returns

Post by Maynard F. Speer »

pkcrafter wrote:Wow, this tread has really gotten some legs. The differing views seem to be mostly between the OP and a handful of posters who found the message incorrect enough to debate it. The great majority of Bogleheads should follow longinvest's message because it is the Boglehead way.

OK, there some Boglehead posters who has taken a long look at valuation models, expected returns, and several other considerations and they are confident enough to incorporate them into portfolio management
Of course Bogle himself (as well as Markowitz - the godfather of "buy and hold" investing) have famously used high valuations to make asset allocation changes ... So I think it's a strange situation where the message has sort of grown legs and run away from those who originally put it out there

(Not that I think elevating particular investors to a deity-like status will necessarily help with returns)
"Economics is a method rather than a doctrine, an apparatus of the mind, a technique of thinking, which helps its possessor to draw correct conclusions." - John Maynard Keynes
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 11:55 pm

Re: The Futility of Predicting Future Returns

Post by lack_ey »

longinvest wrote:2- Major point.

During my 20s, even if I had known a good prediction of the future returns of stocks and bonds, they would have been useless to me for setting a savings rate. I simply had no idea what my future salary would be over the next almost three decades. The only factor that was actionable to decide of a savings rate was my ability to live reasonably below my means and save the excess. Actually, in my early 20s, I went to university, so I had to live at my means and couldn't save money (it was still better than going into debt).

I do not think that a young investor has enough information about his future income in upcoming decades to make any useful use of future return predictions for setting a savings rate. The main deciding factor is the ability to live below one's means (but not like a pauper) and invest the rest.

When I think about it, the only use I see for using predictions, in setting the savings rate of a young investor, is to justify saving less and spending more! But, we all know that common sense should be used, instead, to balance spending and saving.
To me, a prediction of unusually low returns ahead would suggest that saving now is relatively unimportant. If assuming a high (real) rate of return, say 7.2%, then $1 saved today is worth $8 (after adjusting for inflation) in 30 years. Nice! If the real rate of return is more like 2%, then $1 saved today is worth close to $1.80 (inflation adjusted) in 30 years. I'll take it, but bummer.

This would imply that focusing on education (say, spending time in graduate or professional school, earning peanuts and not being able to get in early savings), career development, and higher earnings potential down the road is relatively better and favored. If you're not going to get much compounding, you better be able to save more.

Obviously I don't recommend anybody changing life plans just to chase savings maximization, but you get the point.


Based on just back-of-the-envelope calculations and trivial spreadsheet math, William Bernstein uses some return assumptions to say young people (say 25 and working 40 more years) should save at least about 15% of income here. The reason it doesn't say at least about 12% or at least about 20% is because of the return assumptions and other stipulations about income, including SS support. Actually, it might say a little bit more than 15% if you rerun the numbers today.

It may well turn out that 15% is not enough, or too much, because that's how life works. That's why you update information and estimates over time. Personally, I wouldn't be comfortable with saving only 15%.
User avatar
SimpleGift
Posts: 3869
Joined: Tue Feb 08, 2011 3:45 pm
Location: Central Oregon

Re: The Futility of Predicting Future Returns

Post by SimpleGift »

longinvest wrote:During my 20s, even if I had known a good prediction of the future returns of stocks and bonds, they would have been useless to me for setting a savings rate. I simply had no idea what my future salary would be over the next almost three decades. The only factor that was actionable to decide of a savings rate was my ability to live reasonably below my means and save the excess.
This is certainly understandable and, in your personal situation, a reasonable response in the face of uncertainty.

However, for investors in general, with both stock and bond prices near historical highs, and expected portfolio returns near historical lows, it would also be reasonable to consider planning on saving more, working longer, reducing investment expenses, etc. And for retirees, there's a motivation to carefully consider their withdrawal rates and their cost of living decisions.

At the very least, if folks want to ignore expected returns entirely, I hope they'll not fall back on using historical average asset returns over the past 50 or 100 years to guide their long-range financial planning for the future!
pkcrafter
Posts: 14823
Joined: Sun Mar 04, 2007 12:19 pm
Location: CA
Contact:

Re: The Futility of Predicting Future Returns

Post by pkcrafter »

Maynard wrote:
Of course Bogle himself (as well as Markowitz - the godfather of "buy and hold" investing) have famously used high valuations to make asset allocation changes ... So I think it's a strange situation where the message has sort of grown legs and run away from those who originally put it out there
Point taken, but I won't get too excited. I'll stick with the do nothing strategy and continue to recommend it to new Bogleheads. Valuation strategies aren't a fool proof answer, are they? Anyway, the whole point of my posting is to support Longinvest's message. If you simply ride the market with an asset allocation that is reasonable, you will get your "fair share" and that ain't bad as it beats 80% of everything else over a 20 year period.

Paul
When times are good, investors tend to forget about risk and focus on opportunity. When times are bad, investors tend to forget about opportunity and focus on risk.
Post Reply