Static vs Dynamic Asset Allocation; Victor Meets the Boglehead

[Leesbro63]

I found this interesting article online today and it's relevant here because the authors went to the most recent Bogleheads conference. It's also relevant because they discuss static vs dynamic asset allocation based on expected returns. I'm sure static vs dynamic allocation has been discussed here before, somewhere. At the very least it's interesting to see that the authors attended Bogleheads and what they learned. I posted the beginning of the article and a link to the whole thing, here:

"“Bogleheads” are DIY investors who are passionate about index investing. They gather each year to share ideas about sensible investing, and to celebrate the life and contributions of John Bogle, the founder of Vanguard and arguably the person who has done more than anyone to improve investor welfare. Victor was very pleased to attend their recent annual conference in Minneapolis, and do a Q&A session with Morningstar’s Christine Benz.

There were also about a dozen authors of excellent personal finance books and blogs who gave presentations, including Christine Benz, Rick Ferri, William Bernstein, Allan Roth, Mike Piper, Jackie Cummings Koski, Karsten Jeske and Sarah-Catherine Gutierrez. Victor thoroughly enjoyed the experience, and hopes he’ll be invited back to next year’s conference in Austin!

We agree 100% with almost everything discussed over the course of the three-day conference. However, one area where we noticed our opinions diverge from the Boglehead consensus view was on asset allocation..."


https://www.advisorperspectives.com/art ... -boglehead

[ScubaHogg]

It was a good article. I didn’t realize they were at the conference. Made me disappointed that I didn’t go. I very much enjoyed their book

[Trance]

At the end of the year 2000, the cyclically-adjusted earnings yield of U.S. equities was 2.9% and U.S. inflation protected bonds (TIPS) had a real yield of 3.75%. At the end of 2010, the earnings yield of U.S. equities was 6% and TIPS offered a 1% real yield. If you believe, as we do, that the earnings yield of the equity market is a decent estimate of its long-term real return, then you would not have wanted the same asset allocation at the end of 2000 as you had at the end of 2010. And you would have been justified in owning less equities and more TIPS in 2000, and more equities and less TIPS in 2010. Over the first decade of this century, U.S. equities under-performed 10-year maturity TIPS by over 4% pa, while in the second decade, it was the other way around, with equities outperforming TIPS by 10% pa.

An investor who kept 60% in U.S. stocks and 40% in bonds over the two decades enjoyed a compound return of 7.1%, while an investor who was 30%/70% in stocks/bonds for the first ten years, and then 90%/10% in stocks/bonds for the next ten years – for an average exposure of 60/40 – would have earned a compound return of 9.2%, 2.1% higher, with roughly the same risk. The dynamic asset allocator’s realized Sharpe ratio would have been 27% higher than the Sharpe ratio of the 60/40 static weight investor.

I can understand a bit this angle. A real tips yield of 3.75% sounds genuinely divine. I don't thinks market timing per say to purchase more bonds at favorable yields.

[steve r]

From article:

An investor who kept 60% in U.S. stocks and 40% in bonds over the two decades enjoyed a compound return of 7.1%, while an investor who was 30%/70% in stocks/bonds for the first ten years, and then 90%/10% in stocks/bonds for the next ten years – for an average exposure of 60/40 – would have earned a compound return of 9.2%, 2.1% higher, with roughly the same risk. The dynamic asset allocator’s realized Sharpe ratio would have been 27% higher than the Sharpe ratio of the 60/40 static weight investor.

OK, but how would one possibly know this 20 years ago. Not to mention the minefield of behavioral pitfalls when things do not go so smoothly.

Potential for improvement with dynamic AA ... surely.
Assurance that it will work out ... surely not.

[IDpilot]

I found this interesting article online today and it's relevant here because the authors went to the most recent Bogleheads conference. It's also relevant because they discuss static vs dynamic asset allocation based on expected returns. I'm sure static vs dynamic allocation has been discussed here before, somewhere. At the very least it's interesting to see that the authors attended Bogleheads and what they learned. I posted the beginning of the article and a link to the whole thing, here:

"“Bogleheads” are DIY investors who are passionate about index investing. They gather each year to share ideas about sensible investing, and to celebrate the life and contributions of John Bogle, the founder of Vanguard and arguably the person who has done more than anyone to improve investor welfare. Victor was very pleased to attend their recent annual conference in Minneapolis, and do a Q&A session with Morningstar’s Christine Benz.

There were also about a dozen authors of excellent personal finance books and blogs who gave presentations, including Christine Benz, Rick Ferri, William Bernstein, Allan Roth, Mike Piper, Jackie Cummings Koski, Karsten Jeske and Sarah-Catherine Gutierrez. Victor thoroughly enjoyed the experience, and hopes he’ll be invited back to next year’s conference in Austin!

We agree 100% with almost everything discussed over the course of the three-day conference. However, one area where we noticed our opinions diverge from the Boglehead consensus view was on asset allocation..."


https://www.advisorperspectives.com/art ... -boglehead

Victor Haghani is one of the two authors of The Missing Billionaires in which they present their dynamic asset allocation approached based on the Merton Share. This book has been discussed extensively in other Boglehead threads about that book.

[greg24]

An investor who kept 60% in U.S. stocks and 40% in bonds over the two decades enjoyed a compound return of 7.1%, while an investor who was 30%/70% in stocks/bonds for the first ten years, and then 90%/10% in stocks/bonds for the next ten years – for an average exposure of 60/40

LOLOLOLOLOL. Good luck with that.

[KlangFool]

From article:

An investor who kept 60% in U.S. stocks and 40% in bonds over the two decades enjoyed a compound return of 7.1%, while an investor who was 30%/70% in stocks/bonds for the first ten years, and then 90%/10% in stocks/bonds for the next ten years – for an average exposure of 60/40 – would have earned a compound return of 9.2%, 2.1% higher, with roughly the same risk. The dynamic asset allocator’s realized Sharpe ratio would have been 27% higher than the Sharpe ratio of the 60/40 static weight investor.

OK, but how would one possibly know this 20 years ago. Not to mention the minefield of behavioral pitfalls when things do not go so smoothly.

Potential for improvement with dynamic AA ... surely.
Assurance that it will work out ... surely not.

In summary, if you are lucky and you time it perfectly, you gain 2.1% more per year over that 20 years. I know that I am not lucky and I am not that smart. So, I will stay with static asset allocation.

KlangFool

[RetiredAL]

Not to mention the minefield of behavioral pitfalls when things do not go so smoothly.

IMO, investor's who dabble a lot often have a tendency to say 'those mines won't get me'.

[j.click]

The only thing of value would be to inform us what the next 20yrs will look like, and of course no one can.....

[WoodSpinner]

OP,

Thanks for pointing out an interesting article ….

One aspect that was missing from the discussion is how your Personal Financial Circumstances, Goals and Priorities should affect your AA (Dynamic or Static) over time.

For me, this is a key question — Intuitively it makes sense to shift your AA as your circumstances change since it directly affects your Need, Ability, and Willingness to take on risk. The downside of course is having the discipline to regulate the decision process for making and implementing these changes.

FWIW, I am considering tying my Fixed Income investments to X-Years of Cash Flow Requirements rather than a static figure.

Thoughts?

WoodSpinner

[Watty]

At the end of the year 2000, the cyclically-adjusted earnings yield of U.S. equities was 2.9% and U.S. inflation protected bonds (TIPS) had a real yield of 3.75%. At the end of 2010, the earnings yield of U.S. equities was 6% and TIPS offered a 1% real yield. If you believe, as we do, that the earnings yield of the equity market is a decent estimate of its long-term real return, then you would not have wanted the same asset allocation at the end of 2000 as you had at the end of 2010. And you would have been justified in owning less equities and more TIPS in 2000, and more equities and less TIPS in 2010. Over the first decade of this century, U.S. equities under-performed 10-year maturity TIPS by over 4% pa, while in the second decade, it was the other way around, with equities outperforming TIPS by 10% pa.

An investor who kept 60% in U.S. stocks and 40% in bonds over the two decades enjoyed a compound return of 7.1%, while an investor who was 30%/70% in stocks/bonds for the first ten years, and then 90%/10% in stocks/bonds for the next ten years – for an average exposure of 60/40 – would have earned a compound return of 9.2%, 2.1% higher, with roughly the same risk. The dynamic asset allocator’s realized Sharpe ratio would have been 27% higher than the Sharpe ratio of the 60/40 static weight investor.

I can understand a bit this angle. A real tips yield of 3.75% sounds genuinely divine. I don't thinks market timing per say to purchase more bonds at favorable yields.

Talk about cherry picking dates. 2000 was the beginning of the Dot Com bust so it is basically saying if you underweighted stocks during the Dot Com bust then increased them later that you would do better.

[Gaston]

OK, but how would one possibly know this 20 years ago. Not to mention the minefield of behavioral pitfalls when things do not go so smoothly.

I think you hit the nail on the head. The authors of the article cite an example of how you could have done better had you adopted a dynamic allocation. But they make no mention of how you could have done a lot worse with the wrong dynamic allocation. How convenient.

I also disagree with their point that the tax consequences of dynamic allocations "can be mitigated through tax-loss harvesting and tax-aware rebalancing". I am an avid tax-loss harvester. But the last time I was able to harvest a tax loss of any consequence was in 2020 during the Covid downturn. And before that, it was in 2008 and 2009 during the GFC. If you hold a largely Boglehead portfolio with broad-based equity index funds, the reality is that opportunities to tax loss harvest are few and far between. One can take advantage of the opportunities when they occur, but one cannot plan them.

[KlangFool]

OP,

Thanks for pointing out an interesting article ….

One aspect that was missing from the discussion is how your Personal Financial Circumstances, Goals and Priorities should affect your AA (Dynamic or Static) over time.

For me, this is a key question — Intuitively it makes sense to shift your AA as your circumstances change since it directly affects your Need, Ability, and Willingness to take on risk. The downside of course is having the discipline to regulate the decision process for making and implementing these changes.

FWIW, I am considering tying my Fixed Income investments to X-Years of Cash Flow Requirements rather than a static figure.

Thoughts?

WoodSpinner

I would like to understand what do you mean by X years. For example, do you always keep 10 years in retirement regardless how old you are?

KlangFool

[Ben Mathew]

Great article. I think Haghani hit the nail on the head with this reason for the popularity of the static asset allocation:

• "We suspect that, in many cases where an investor’s estimated returns and risk are quite constant over time, what’s implicitly happening is that those estimates are primarily being anchored to very long-term historical returns. Long- term historical returns are pretty constant over time, and so an asset allocation using them as inputs will be pretty static too."

I see this happen a lot. People pick an asset allocation based on historical backtesting, and this implicit expected return assumption gets baked into the analysis.

Another common pathway I've seen is the implicit assumption that stocks will definitely do better than bonds over long horizons. This leads people to think that they should hold as much as stocks as they can for long term goals like retirement, constrained primarily by the ability to stick to the plan during a crash and "sleep well at night." This focuses exclusively on short term stock risk and behavioral concerns, and assumes away long term stock risk.

It's also worth emphasizing that a static asset allocation is not a passive approach that assumes market efficiency. It involves actively trading to maintain the same allocation. Everybody in the market cannot hold a static asset allocation because the market cap of stocks and bonds fluctuate. If stocks fall and you want to maintain your original asset allocation, you'd have to buy stock from someone else. So the other person has to shift towards even more bonds. It's hard to justify this if markets are efficient and stocks and bonds are priced correctly. A truly passive approach involves holding market cap weighted allocations—effectively buy and hold. Nobody needs to trade as markets rise and fall. Bill Sharpe has suggested such a portfolio as an efficient portfolio that requires little trading to maintain. Static asset allocation is quite different from that. It's not passive and not consistent with market efficiency.

[White Coat Investor]

I found this interesting article online today and it's relevant here because the authors went to the most recent Bogleheads conference. It's also relevant because they discuss static vs dynamic asset allocation based on expected returns. I'm sure static vs dynamic allocation has been discussed here before, somewhere. At the very least it's interesting to see that the authors attended Bogleheads and what they learned. I posted the beginning of the article and a link to the whole thing, here:

"“Bogleheads” are DIY investors who are passionate about index investing. They gather each year to share ideas about sensible investing, and to celebrate the life and contributions of John Bogle, the founder of Vanguard and arguably the person who has done more than anyone to improve investor welfare. Victor was very pleased to attend their recent annual conference in Minneapolis, and do a Q&A session with Morningstar’s Christine Benz.

There were also about a dozen authors of excellent personal finance books and blogs who gave presentations, including Christine Benz, Rick Ferri, William Bernstein, Allan Roth, Mike Piper, Jackie Cummings Koski, Karsten Jeske and Sarah-Catherine Gutierrez. Victor thoroughly enjoyed the experience, and hopes he’ll be invited back to next year’s conference in Austin!

We agree 100% with almost everything discussed over the course of the three-day conference. However, one area where we noticed our opinions diverge from the Boglehead consensus view was on asset allocation..."


https://www.advisorperspectives.com/art ... -boglehead

Two things going on here.

# 1 They've got to justify to clients why they're doing tactical asset allocation
# 2 They've got to justify to themselves why they're doing tactical asset allocation

Lots of people believe their method is better than a static allocation, but the fact that they're all doing it differently and most are underperforming a know nothing static allocation isn't particularly inspiring. Vanguard had a fund where they tried to do this for a while but shut it down because they sucked at it so bad. Like the rest of us.

Successful, long-term tactical asset allocation is hard, especially after paying the costs including taxes and often advisory fees. Impossible? Probably not but I do think the right thing to do is to assume it is. Kind of like market efficiency. Markets aren't completely efficient, but you should still act like they are.

[er999]

Seems like changing asset allocation is a service they are selling, made clear by the trademark after their preferred description multiple times in the article. Complexity is good for his business — same with the Merton’s share equation with the Greek letters in the formula to make it seem more authoritative.

[assetmix]

I read their book last year, which I enjoyed, and released a podcast episode on it. The whole idea is that our allocation to stocks or other risky assets is a function of the real expected return, volatility, and our risk tolerance.

Whether we are strategic or more dynamic allocators, that shouldn't be controversial. For example, allocating more to TIPS and locking in higher yields made more sense over the last couple of years because real yields reached over 2%. Conversely, three years ago real yields on TIPS were negative so Series I Savings bonds made sense even at a 0% base yield. Perhaps the allocation to TIPS was done as part of a rebalancing away from stocks.

As for expected returns, even strategic allocators should have an understanding of the range of potential returns and the underlying drivers. In other words, how much are we getting paid to take on the potential drawdown risk of stocks compared to bonds.

For example, here are some 10-year expected stock returns and bond returns. The stock returns assume valuations fall to the median and earnings equal their long-term 5-year averages. The bond returns assume interest rates increase 1%.





An allocation to value and bonds appears like a safer bet than an allocation to growth at this juncture.

We don't know what will happen, but we know what yields and valuations are today. We make judgments of relative values and potential payoffs in many areas of our lives. We can do the same as investors.

[Northern Flicker]

The problem is that, for most investments, we do not know the expected return or variance/risk, real or nominal, over any particular time horizon.

Note that consideration of the variance of real return and variance of nominal return over a broad set of time horizons captures a broad range of (and most) risk measures, so this is not equating risk with volatility.

So, yes, you can adjust asset allocation based on changes in estimates of expected return and risk, but without reasonably accurate measures of expected return and risk, the adjustment may increase or decrease alignment with an investor's objectives and risk tolerance.

[assetmix]

The problem is that, for most investments, we do not know the expected return or variance/risk, real or nominal, over any particular time horizon.

Note that consideration of the variance of real return and variance of nominal return over a broad set of time horizons captures a broad range of (and most) risk measures, so this is not equating risk with volatility.

So, yes, you can adjust asset allocation based on changes in estimates of expected return and risk, but without reasonably accurate measures of expected return and risk, the adjustment may increase or decrease alignment with an investor's objectives and risk tolerance.

I agree that volatility is not an intuitive or even a helpful measure of risk for most people. That's why I don't use it. For me, risk is losing money. Risk is financial ruin. If stocks fall 65%, then that could lead to retirement ruin for many retirees.

Still, I don't think we have to be terribly accurate forecasters to make allocation decisions. If I know I can lock in inflation plus 2% on an individual TIPS compared to a 1.3% dividend yield for the S&P 500 at a Shiller P/E of 34, then it is pragmatic to choose more of the former and less of the latter if I would be devastated by a 65% decline in the S&P and it might take more two decades to recover as in the case of Japan.

[nisiprius]

1) I agree that the presence of an ® symbol after Dynamic Index Investing® is an indication that he is not a disinterested party.

2) Everything turns on this:

We often hear people say that our Dynamic Index Investing® approach doesn’t make sense because it is not possible to estimate the expected return and riskiness of stock markets. As we explained earlier in this note, we strongly believe that it is possible to reasonably make those estimates...

File this under "if you can predict something about the stock market, then you can do better by exploiting that prediction than not exploiting it." But can you? We are assured that he "strongly believes" that you can.

However, according to Wikipedia's article on Victor Haghani, "in 1993, he co-founded Long-Term Capital Management with seven other partners." I must assume that he "strongly believed" in LTCM's strategy, too. (Why, yes, I do think participation in LTCM should be held against the experts who participated in it).

3)

...we recognize that dynamic asset allocation may not be appropriate for many investors

OK. I recognize myself as one of those for whom it is not appropriate.

4)

There are many 10-year periods over which dynamic asset allocation would have resulted in a return below and/or a risk above that of a static asset allocation.

He then presents an example a specific time period "when dynamic asset allocation worked well."

But the thing is, the example is from 2000, and Elm Wealth was founded in 2011. He chose an example from a time period before it was founded, and probably before Dynamic Index Investing® had been codified. So it does not represent actual client experience. It probably does not represent an out-of-sample test of Dynamic Index Investing®. It probably represents 20/20 hindsight incorporated within the strategy.

5)

However, we believe that changing your asset allocation over time as the expected excess return and risk of stocks change, is a more logical approach than keeping your allocation constant through time.

I think most of Wall Street believes this. Just as I think most of Wall Street believes "any attempt to evaluate stocks and concentrate on the good ones is more logical than indexing."

[Hydromod]

The Merton's share allocation idea is pretty standard for this kind of thing.

There's no real reason to think that expected returns and volatility are correlated. If you have no information on expected returns, then both the adaptive strategy and the fixed allocation strategy can be described as just weighting the assets proportionally to their risk budget and inversely to their variance. The risk budget is just the fraction of portfolio volatility coming from each asset.

Haghani hit the nail on the head with the frequency of update. Having a fixed allocation is consistent with using very long-term estimates. If you assume that expected returns are fixed, you can end up with the risk budget allocation from long-term volatility and static allocation.

Or you can allocate using the most recent few months of volatility, which is predictive of future volatility to some extent.

I can run backtests from 1962 on a "SPY"/"TLT" or "SPY"/"IEF" (S&P 500 plus long-term or intermediate-term treasuries) using simulated values before inception. I find that over that period, running a backtest with an inverse variance approach and running it with the same average allocation and same rebalancing frequency comes up with similar CAGR and Sharpe for the two approaches (before taxes). The adaptive one pays more taxes, even though the rebalancing frequency is identical. Which one comes out ahead before taxes depends on the particulars of rebalancing.

Where it makes more of a difference is when you are working with leverage. That's when reducing allocations during periods of high volatility helps mitigate the volatility decay and big losses that eat away at returns.

[watchnerd]

From article:

An investor who kept 60% in U.S. stocks and 40% in bonds over the two decades enjoyed a compound return of 7.1%, while an investor who was 30%/70% in stocks/bonds for the first ten years, and then 90%/10% in stocks/bonds for the next ten years – for an average exposure of 60/40 – would have earned a compound return of 9.2%, 2.1% higher, with roughly the same risk. The dynamic asset allocator’s realized Sharpe ratio would have been 27% higher than the Sharpe ratio of the 60/40 static weight investor.

OK, but how would one possibly know this 20 years ago. Not to mention the minefield of behavioral pitfalls when things do not go so smoothly.

Potential for improvement with dynamic AA ... surely.
Assurance that it will work out ... surely not.

Interest rates are observable in the present as as signal and a major clue.

When interest rates are zero or negative, ERP should be higher.

[watchnerd]

A truly passive approach involves holding market cap weighted allocations—effectively buy and hold. Nobody needs to trade as markets rise and fall. Bill Sharpe has suggested such a portfolio as an efficient portfolio that requires little trading to maintain. Static asset allocation is quite different from that. It's not passive and not consistent with market efficiency.

Our risk portfolio is a modified* version of the global market port.

We don't rebalance it.

(*because we have a TIPS LMP ladder, we exclude Treasuries from the risk port, otherwise we'd be massively overweight Treasuries)

[unwitting_gulag]

Another common pathway I've seen is the implicit assumption that stocks will definitely do better than bonds over long horizons. This leads people to think that they should hold as much as stocks as they can for long term goals like retirement, constrained primarily by the ability to stick to the plan during a crash and "sleep well at night." This focuses exclusively on short term stock risk and behavioral concerns, and assumes away long term stock risk.

Is there any data suggesting that over sufficiently long periods (say, 50+ years), stocks have ever underperformed bonds? In other words, the nod towards stocks is a personal statement about not needing the money... maybe not ever, or at least, not over any foreseeable period. If that's so, then "holding as much stocks as they can" sounds reasonable, does it not?

It's also worth emphasizing that a static asset allocation is not a passive approach that assumes market efficiency. It involves actively trading to maintain the same allocation. Everybody in the market cannot hold a static asset allocation because the market cap of stocks and bonds fluctuate. ...

I'd argue that a truly passive approach means doing nothing. Never rebalance! Let the portfolio drift where it may... stocks vs. bonds, US vs. ex-US, large-cap vs. small-cap. Whether rebalancing is itself a form of market-timing, or not... has been a longstanding debate here, has it not? But I think that the more insightful comparison between BH and Hagani takes for the former the passively drifting portfolio, rather than one, where the asset allocation is regularly nudged to stay within some guard rails.

[bobcat2]

We agree 100% with almost everything discussed over the course of the three-day conference. However, one area where we noticed our opinions diverge from the Boglehead consensus view was on asset allocation..."

If you don't agree with someone else on how to do asset allocation, you are nowhere near 100% agreement with the other person on how to do financial planning.

BobK

[retired@50]

1) I agree that the presence of an ® symbol after Dynamic Index Investing® is an indication that he is not a disinterested party.

2) Everything turns on this:

We often hear people say that our Dynamic Index Investing® approach doesn’t make sense because it is not possible to estimate the expected return and riskiness of stock markets. As we explained earlier in this note, we strongly believe that it is possible to reasonably make those estimates...

File this under "if you can predict something about the stock market, then you can do better by exploiting that prediction than not exploiting it." But can you? We are assured that he "strongly believes" that you can.

However, according to Wikipedia's article on Victor Haghani, "in 1993, he co-founded Long-Term Capital Management with seven other partners." I must assume that he "strongly believed" in LTCM's strategy, too. (Why, yes, I do think participation in LTCM should be held against the experts who participated in it).

3)

...we recognize that dynamic asset allocation may not be appropriate for many investors

OK. I recognize myself as one of those for whom it is not appropriate.

4)

There are many 10-year periods over which dynamic asset allocation would have resulted in a return below and/or a risk above that of a static asset allocation.

He then presents an example a specific time period "when dynamic asset allocation worked well."

But the thing is, the example is from 2000, and Elm Wealth was founded in 2011. He chose an example from a time period before it was founded, and probably before Dynamic Index Investing® had been codified. So it does not represent actual client experience. It probably does not represent an out-of-sample test of Dynamic Index Investing®. It probably represents 20/20 hindsight incorporated within the strategy.

5)

However, we believe that changing your asset allocation over time as the expected excess return and risk of stocks change, is a more logical approach than keeping your allocation constant through time.

I think most of Wall Street believes this. Just as I think most of Wall Street believes "any attempt to evaluate stocks and concentrate on the good ones is more logical than indexing."

Wow. A surgical and logical dismantling of each and every statement / argument.

Nice work nisiprius.

Regards,

[nisiprius]

...Is there any data suggesting that over sufficiently long periods (say, 50+ years), stocks have ever underperformed bonds?..

Yes. There has been a 68-year span. (And also a 41-year span).

Rob Arnott: Bonds: Why Bother?



And from no less a luminary than Jeremy Siegel, in Stocks for the Long Run 5/E, although he only said "longer than thirty years" and didn't mention the full length of the period:

In the first four editions of Stocks for the Long Run, I noted that the last 30-year period when the return on long-term bonds beat stocks ended in 1861, at the onset of the U.S. Civil War. That is no longer true.

Finally, McQ (his name in this forum), in his 2021 paper Where Siegel Went Wrong exhibits this table:



Over his long historical period, the chances that stocks beat bonds over a randomly-chosen 50-year period was only 68.4%, meaning there was over a 31% chances that bonds beat stocks.

[Northern Flicker]

The problem is that, for most investments, we do not know the expected return or variance/risk, real or nominal, over any particular time horizon.

Note that consideration of the variance of real return and variance of nominal return over a broad set of time horizons captures a broad range of (and most) risk measures, so this is not equating risk with volatility.

So, yes, you can adjust asset allocation based on changes in estimates of expected return and risk, but without reasonably accurate measures of expected return and risk, the adjustment may increase or decrease alignment with an investor's objectives and risk tolerance.

I agree that volatility is not an intuitive or even a helpful measure of risk for most people. That's why I don't use it. For me, risk is losing money. Risk is financial ruin. If stocks fall 65%, then that could lead to retirement ruin for many retirees.
...
If I know I can lock in inflation plus 2% on an individual TIPS compared to a 1.3% dividend yield for the S&P 500 at a Schiller P/E of 34, then it is pragmatic to choose more of the former and less of the latter if I would be devastated by a 65% decline in the S&P and it might take more two decades to recover as in the case of Japan.

Your belief in dynamic asset allocation is leading you astray there. If you would be devastated by a 65% or whatever decline in your portfolio, then you should be holding a portfolio with a very low to zero likelihood of such a decline regardless of the dividend yield or the P/E of the S&P500 or TIPS yields.

[Ben Mathew]

Is there any data suggesting that over sufficiently long periods (say, 50+ years), stocks have ever underperformed bonds? In other words, the nod towards stocks is a personal statement about not needing the money... maybe not ever, or at least, not over any foreseeable period. If that's so, then "holding as much stocks as they can" sounds reasonable, does it not?

From another post that I wrote about this:

• In Shiller's data that I looked at here, stocks always outperformed bonds over 30 year periods over the last 150 years. So can we conclude that stocks will always beat bonds over 30 year horizons going forward? No. Because if that were true, then there is an arbitrage opportunity. You (or banks or funds or corporations or any other institution) can issue 30 year non-callable bonds, invest the money in stocks, and pocket the difference. Infinite leverage to get infinite profits. Clearly the market does not believe this. And neither should we. History was just one path. At the very least, we should be doing Monte Carlo simulations that randomly samples with replacement from the past data. That itself opens the door to stocks doing worse than bonds over long horizons. I would go one step further and lower the expected return of the distribution because I believe that forward looking expected returns today are lower than raw historical returns. But even without that, just random sampling from the past historical distribution will give you a non-zero probability of stocks underperforming bonds over any horizon. It's a small probability, but not zero. And it's a small probability of a very bad event. Hence the premium for holding equity, even over long horizons. Otherwise, the market prices of assets just does not make sense.
  

But I think that the more insightful comparison between BH and Hagani takes for the former the passively drifting portfolio, rather than one, where the asset allocation is regularly nudged to stay within some guard rails.

What I've seen typically recommended on the forum is rebalancing back to the original asset allocation, not letting the portfolio passively drift.

[bobcat2]

It's interesting that they base everything on the Merton share, which Bob Merton developed when he was a grad student in 1969. But these days in designing retirement portfolios Merton himself does not rely on the Merton share. The Merton share is an interesting theoretical first pass on how to do asset allocation, but using it literally to do financial planning is probably a stretch.

These days this is the way Bob Merton approaches retirement planning.

…a liability-driven investment strategy will be implemented with the aim to improve the likelihood of achieving their desired [retirement] income. …


[The] solution that I am proposing here first determines the allocation of the member’s total assets to an inflation-linked duration-matched fixed-income portfolio that will be required to achieve the member’s conservative income target with a 96 per cent estimated probability. It then uses the remaining assets to improve the estimated probability of achieving the desired income target. Investment risk is measured in terms of uncertainty about income in retirement and not wealth. The strategies are explicitly designed to manage inflation and interest rate risks as well as asset price risks. Investment risk will only be taken when it is needed to reach the member’s income target.

Applying life-cycle economics - Robert Merton
https://doczz.net/doc/8871437/applying- ... -economics

BobK

[bobcat2]

I see Shiller's name has been brought up in this discussion. I note that Shiller's CAPE today is higher than at any other time going back to 1871 other than late 1999 and early 2000.

Link to graph of CAPE - https://www.multpl.com/shiller-pe

BobK

[assetmix]

I see Shiller's name has been brought up in this discussion. I note that Shiller's CAPE today is higher than at any other time going back to 1871 other than late 1999 and early 2000.

Link to graph of CAPE - https://www.multpl.com/shiller-pe

BobK

Almost two standard deviations more expensive than average using inflation-adjusted earnings data going back to 1972.

[Peter G]

I'd like to challenge and raise a question about the use of 'odds' in this table, where I think probability is the measure actually being used. Odds and probability are not equivalent, as seen in the formula 'odds = probability / (1- probability)'.
In this 2021 paper we can read the explanation for the calculations shown in the Table 1 extract. It says: 'Annualised returns were calculated for stocks and bonds and then a count was taken of the number of rolls where the stock returns exceeded the bond return. The percentage stated is that count divided by the total number of rolls'. That seems to me to be a definition of probability. Odds, by contast, is the number of occurrences divided by the number of non-occurrences. Thus, if an event occurs 50% of the time, the probability of it occurring is 50%, but the odds is 100%. I think. Would welcome correcting me.

[ScubaHogg]

It’s almost as if the authors haven’t written a book addressing most of these objections…

[ScubaHogg]

Talk about cherry picking dates. 2000 was the beginning of the Dot Com bust so it is basically saying if you underweighted stocks during the Dot Com bust then increased them later that you would do better.

Well no, it’s not “cherry picking.” That period was exactly when the real return of TIPS was sky high and the earnings yields of stocks was in the gutter. So the risk premium was stocks had utterly tanked.

It’s exactly when the logic of moving your AA around, whether you personally want to or not, would have you more heavily in the guaranteed return of TIPS.

[Dottie57]

From article:


OK, but how would one possibly know this 20 years ago. Not to mention the minefield of behavioral pitfalls when things do not go so smoothly.

Potential for improvement with dynamic AA ... surely.
Assurance that it will work out ... surely not.

In summary, if you are lucky and you time it perfectly, you gain 2.1% more per year over that 20 years. I know that I am not lucky and I am not that smart. So, I will stay with static asset allocation.

KlangFool

I am sorry but this seems like market timing in fancy clothes. The authors say not. Buuuutttttt..

[ScubaHogg]

The problem is that, for most investments, we do not know the expected return or variance/risk, real or nominal, over any particular time horizon.

Note that consideration of the variance of real return and variance of nominal return over a broad set of time horizons captures a broad range of (and most) risk measures, so this is not equating risk with volatility.

So, yes, you can adjust asset allocation based on changes in estimates of expected return and risk, but without reasonably accurate measures of expected return and risk, the adjustment may increase or decrease alignment with an investor's objectives and risk tolerance.

But of course, as the authors point out, you have to, either explicitly or implicitly, make some assumptions about an expected return and volatility. If you don’t you really shouldn’t be investing in risky assets at all.

[nisiprius]

(image)

I'd like to challenge and raise a question about the use of 'odds' in this table, where I think probability is the measure actually being used. Odds and probability are not equivalent, as seen in the formula 'odds = probability / (1- probability)'.
In this 2021 paper we can read the explanation for the calculations shown in the Table 1 extract. It says: 'Annualised returns were calculated for stocks and bonds and then a count was taken of the number of rolls where the stock returns exceeded the bond return. The percentage stated is that count divided by the total number of rolls'. That seems to me to be a definition of probability. Odds, by contast, is the number of occurrences divided by the number of non-occurrences. Thus, if an event occurs 50% of the time, the probability of it occurring is 50%, but the odds is 100%. I think. Would welcome correcting me.

I think you're correct. I interpreted the numbers as "probabilities."

Except that I'm not sure how odds are stated these days, Bayesian statistics only coming into vogue after I took Statistics 101. I've never seen odds expressed as a percentage. In my day, anyway, odds were usually stated with a notation like A:B, which is illustrated in Wikipedia by a picture:

[ScubaHogg]

I am sorry but this seems like market timing in fancy clothes. The authors say not. Buuuutttttt..

I guess it depends on how you define “market timing.” If you define it as “ever changing your asset allocation no matter what happens” I reckon it is.

If you define it more reasonably as something closer to, “changing your asset allocation to benefit from short term asset mispricings” then no, it’s not.

The question for the group would be, is there any TIPS real yield that would lead you to completely abandon equities? 4%? 10%? 50%?

If the answer is “yes”, then you are agreeing in principle with their approach. If the answer is “no”, well…I don’t know. That just seems odd to me for a group that is constantly and tediously asking questions about SWRs

[Northern Flicker]

The problem is that, for most investments, we do not know the expected return or variance/risk, real or nominal, over any particular time horizon.

Note that consideration of the variance of real return and variance of nominal return over a broad set of time horizons captures a broad range of (and most) risk measures, so this is not equating risk with volatility.

So, yes, you can adjust asset allocation based on changes in estimates of expected return and risk, but without reasonably accurate measures of expected return and risk, the adjustment may increase or decrease alignment with an investor's objectives and risk tolerance.

But of course, as the authors point out, you have to, either explicitly or implicitly, make some assumptions about an expected return and volatility. If you don’t you really shouldn’t be investing in risky assets at all.

But 65% and similar drops still generally are the result of unpredictable, future adverse events.

[ScubaHogg]

But 65% and similar drops still generally are the result of unpredictable, future adverse events.

True, but even then we are implicitly assuming those don’t occur too often. If our assumption was that those were yearly I doubt anyone on here would hold many equities

We are also making assumptions, either explicitly or implicitly, that company earnings are somehow tied to return. Otherwise we’d have zero problem with Greater Fool investing (bitcoin, etc.) or negative expected return investing (lottery tickets)

[watchnerd]

The problem is that, for most investments, we do not know the expected return or variance/risk, real or nominal, over any particular time horizon.

Note that consideration of the variance of real return and variance of nominal return over a broad set of time horizons captures a broad range of (and most) risk measures, so this is not equating risk with volatility.

So, yes, you can adjust asset allocation based on changes in estimates of expected return and risk, but without reasonably accurate measures of expected return and risk, the adjustment may increase or decrease alignment with an investor's objectives and risk tolerance.

But of course, as the authors point out, you have to, either explicitly or implicitly, make some assumptions about an expected return and volatility. If you don’t you really shouldn’t be investing in risky assets at all.

100%.

Playing around with Black Litterman MVO makes this clear, even if the assumptions are just based on using historical averages.

FWIW, my risk port has some assets in the 25-30% Std Dev historical range which makes for some wild MVO optimal port swings, even at small percentages.

[ScubaHogg]

Two things going on here.

# 1 They've got to justify to clients why they're doing tactical asset allocation
# 2 They've got to justify to themselves why they're doing tactical asset allocation

Lots of people believe their method is better than a static allocation, but the fact that they're all doing it differently and most are underperforming a know nothing static allocation isn't particularly inspiring. Vanguard had a fund where they tried to do this for a while but shut it down because they sucked at it so bad. Like the rest of us.

Successful, long-term tactical asset allocation is hard, especially after paying the costs including taxes and often advisory fees. Impossible? Probably not but I do think the right thing to do is to assume it is. Kind of like market efficiency. Markets aren't completely efficient, but you should still act like they are.

James, question. If I understand correctly you do a fair amount of direct real estate investing. If that is true, do you ever concern yourself with the price you are paying for a property relative to its rent? Or do you just buy at any asking price?

[watchnerd]

I am sorry but this seems like market timing in fancy clothes. The authors say not. Buuuutttttt..

I guess it depends on how you define “market timing.” If you define it as “ever changing your asset allocation no matter what happens” I reckon it is.

If you define it more reasonably as something closer to, “changing your asset allocation to benefit from short term asset mispricings” then no, it’s not.

The question for the group would be, is there any TIPS real yield that would lead you to completely abandon equities? 4%? 10%? 50%?

If the answer is “yes”, then you are agreeing in principle with their approach. If the answer is “no”, well…I don’t know. That just seems odd to me for a group that is constantly and tediously asking questions about SWRs

Yes, of course.

TIPS real yields at >4% would make me question the need for equities unless they were absolutely in the gutter in terms of price to earnings and expected future returns.

I could easily meet all my financial objects at >4% real yield. And, in fact, my current estimate of my total port (stocks, bonds, everything) is a real return for 10 years of 4.25%.

That's only a 2.28% premium over current 10 YR TIPS.

To say otherwise would be to ignore utility functions in favor of dogma.

[ScubaHogg]

Thought worth sharing this quote from Boglehead approved author William Bernstein

…there are risky assets, there are riskless assets, and there is an exchange rate between them.
- Rational Expectations

He understood that the relationship between equities and riskless assets varies over time, which is essentially the same point Elm Wealth is making.

Whether you want to do anything about that is up to you, but it’s not a crazy idea

[ScubaHogg]

Yes, of course.

TIPS real yields at >4% would make me question the need for equities unless they were absolutely in the gutter in terms of price to earnings and expected future returns.

To say otherwise would be to ignore utility functions in favor of dogma.

[watchnerd]

Thought worth sharing this quote from Boglehead approved author William Bernstein

…there are risky assets, there are riskless assets, and there is an exchange rate between them.
- Rational Expectations

He understood that the relationship between equities and riskless assets varies over time, which is essentially the same point Elm Wealth is making.

Whether you want to do anything about that is up to you, but it’s not a crazy idea

I would go even further and say if one doesn't have a rough understanding of this relationship, one doesn't know where on the risk curve one is shooting for.

Holding 60% stocks in 2014, relative to price to earnings and risk free rates, is not the same spot on the risk spectrum as 2024.

[IDpilot]

I'd like to challenge and raise a question about the use of 'odds' in this table, where I think probability is the measure actually being used. Odds and probability are not equivalent, as seen in the formula 'odds = probability / (1- probability)'.
In this 2021 paper we can read the explanation for the calculations shown in the Table 1 extract. It says: 'Annualised returns were calculated for stocks and bonds and then a count was taken of the number of rolls where the stock returns exceeded the bond return. The percentage stated is that count divided by the total number of rolls'. That seems to me to be a definition of probability. Odds, by contast, is the number of occurrences divided by the number of non-occurrences. Thus, if an event occurs 50% of the time, the probability of it occurring is 50%, but the odds is 100%. I think. Would welcome correcting me.

I think you're correct. I interpreted the numbers as "probabilities."

Except that I'm not sure how odds are stated these days, Bayesian statistics only coming into vogue after I took Statistics 101. I've never seen odds expressed as a percentage. In my day, anyway, odds were usually stated with a notation like A:B, which is illustrated in Wikipedia by a picture:

Probability is the likelihood of an event happening expressed as a fraction or a percent between 0 and 1. It is calculated by comparing the number of favorable outcomes to the total number of outcomes.

Odds are the likelihood of event happening compared to the likelihood of it not happening expressed as a fraction or a ratio.

Thus, an event with a probability of occurring of 50% has the odds of occurring of 50/50 not 100%

[KEotSK66]

There's no real reason to think that expected returns and volatility are correlated. If you have no information on expected returns, then both the adaptive strategy and the fixed allocation strategy can be described as just weighting the assets proportionally to their risk budget and inversely to their variance. The risk budget is just the fraction of portfolio volatility coming from each asset.

re the statement in red...There is a real reason: volatility regimes, which link higher asset class return to low volatility and lower asset class return to high volatility. Volatility regimes are a big part of risk-parity, risk-balancing, risk-allocating, etc.

re the statements in blue...Barring an unforeseen disaster I think some managers do have information on expected returns, don't some managers look at fundamentals such earnings prospects? I would think the market wouldn't make such a big deal about earnings reports, such as XYZ Corporation missing its earnings target/expectation by a small amount, if no information was available.

Not looking to argue so don't take my comments as criticism, good luck.

[White Coat Investor]

Two things going on here.

# 1 They've got to justify to clients why they're doing tactical asset allocation
# 2 They've got to justify to themselves why they're doing tactical asset allocation

Lots of people believe their method is better than a static allocation, but the fact that they're all doing it differently and most are underperforming a know nothing static allocation isn't particularly inspiring. Vanguard had a fund where they tried to do this for a while but shut it down because they sucked at it so bad. Like the rest of us.

Successful, long-term tactical asset allocation is hard, especially after paying the costs including taxes and often advisory fees. Impossible? Probably not but I do think the right thing to do is to assume it is. Kind of like market efficiency. Markets aren't completely efficient, but you should still act like they are.

James, question. If I understand correctly you do a fair amount of direct real estate investing. If that is true, do you ever concern yourself with the price you are paying for a property relative to its rent? Or do you just buy at any asking price?

I don't do direct investing, only passively. Obviously the less you pay the better your return though. I don't think that's a hard point to acknowledge. All else equal, better to pay $10 for a dollar of earnings/rent/interest than $15.

That doesn't make tactical asset allocation any less challenging though.