Rate of Return on Everything: A 150-year History

[SeattleCPA]

Apologies if someone has already pointed to this link and I missed it. But there is a really interesting research paper available at link below that talks about rates of return for 16 different countries for equities, housing, and then riskless assets.

Really interesting stuff--maybe especially for real estate investors since it shows historically unleveraged real estate has produced the same return as equities but with about half the risk:

https://www.frbsf.org/economic-research/files/wp2017-25.pdf

The other thing, really interesting to anybody who tries to employ modern portfolio theory to their investing, is that it looks like housing and equities are not very correlated.

[HawkeyeNFO]

123 pages....I'll wait until one night when I can't go to sleep.

[tralfamadorian]

123 pages....I'll wait until one night when I can't go to sleep.

Compared to the GOP tax bill, this is akin to The Count of Monte Cristo.

Personally, I found it really fascinating. Thanks for posting it @SeattleCPA !

I'm surprised this thread hasn't gotten more replies. This kind of data is usually like catnip to the more detailed oriented MMM folks...

In particular interest to me is Chart 7 on pg. 25, which separates out capital gain for real estate, rental income, capital gain for stocks and dividend income along with the standard deviation for each and the percentage of the gain for each asset class allocated to each.

The standard deviation of 0.75 for yearly rental income of 5.33% specifically is interesting as it could be the basis for an argument that a diversified real estate portfolio could be a suitable, if not preferable, bond alternative.

The second area that grabbed me was Figure 9 on pg. 30 on the average cost of running a housing investment as both a percentage of house value and percentage of rental income. The data was surprisingly consistent (looking at the US) over a 100 year time frame.

[SeattleCPA]

123 pages....I'll wait until one night when I can't go to sleep.

Compared to the GOP tax bill, this is akin to The Count of Monte Cristo.

Personally, I found it really fascinating. Thanks for posting it @SeattleCPA !

I'm surprised this thread hasn't gotten more replies. This kind of data is usually like catnip to the more detailed oriented MMM folks...

In particular interest to me is Chart 7 on pg. 25, which separates out capital gain for real estate, rental income, capital gain for stocks and dividend income along with the standard deviation for each and the percentage of the gain for each asset class allocated to each.

The standard deviation of 0.75 for yearly rental income of 5.33% specifically is interesting as it could be the basis for an argument that a diversified real estate portfolio could be a suitable, if not preferable, bond alternative.

The second area that grabbed me was Figure 9 on pg. 30 on the average cost of running a housing investment as both a percentage of house value and percentage of rental income. The data was surprisingly consistent (looking at the US) over a 100 year time frame.

The study is really interesting. Per my wife, I had a strange smile on my face the entire time I was reading it.

P.S. High point for me was seeing how poorly correlated equities and housing are.... MPT here I come.

[tct]

Interesting. Thanks for sharing.

[SwordGuy]

It's going to take awhile for my tired brain to digest - but this looks to have some real nuggets of gold in it.

[SeattleCPA]

It's going to take awhile for my tired brain to digest - but this looks to have some real nuggets of gold in it.

On some other forums, people are pooh-poohing it. But you've got roughly 150 years of investment returns for 16 countries. There is lots of interesting data in there.

It's not quite like we've now got 2400 years of investment returns (150 years in 16 different developed economies)... but it's a lot better than just 150 years.

For example: It's very interesting to consider their point that riskless asset returns (e.g., treasuries) are not low by historical standards but basically about where you'd expect looking at the long run.

[Mr. Boh]

Really interesting. Great link SeattleCPA.

I haven't read the whole thing but there are a couple of rules of thumb that might need to be reexamined in light of the conclusions reached in this paper. The first is the 1% rule. If real estate has the expected returns of stocks while being uncorrelated and half the risk, it seems to me that you wouldn't need to get anywhere near 1% on a real estate investment for it to make sense within a broader portfolio. The other thing that comes to mind is the perennial rent vs. own debate. The returns on real estate would indicate that one would be better to buy a house rather than renting.

Personally this research has given me some peace of mind. I had a short-lived thread about how to allocate rental properties within a larger portfolio.
https://forum.mrmoneymustache.com/investor-alley/asset-allocation-are-rental-houses-more-like-stocks-or-bonds/msg1746752/#msg1746752
This paper seems to provide an answer to my question.

On some other forums, people are pooh-poohing it. But you've got roughly 150 years of investment returns for 16 countries. There is lots of interesting data in there.

I'm curious, which forums are you referring to and what is the general criticism?

[SeattleCPA]

Really interesting. Great link SeattleCPA.

I haven't read the whole thing but there are a couple of rules of thumb that might need to be reexamined in light of the conclusions reached in this paper. The first is the 1% rule. If real estate has the expected returns of stocks while being uncorrelated and half the risk, it seems to me that you wouldn't need to get anywhere near 1% on a real estate investment for it to make sense within a broader portfolio. The other thing that comes to mind is the perennial rent vs. own debate. The returns on real estate would indicate that one would be better to buy a house rather than renting.

Personally this research has given me some peace of mind. I had a short-lived thread about how to allocate rental properties within a larger portfolio.
https://forum.mrmoneymustache.com/investor-alley/asset-allocation-are-rental-houses-more-like-stocks-or-bonds/msg1746752/#msg1746752
This paper seems to provide an answer to my question.

On some other forums, people are pooh-poohing it. But you've got roughly 150 years of investment returns for 16 countries. There is lots of interesting data in there.

I'm curious, which forums are you referring to and what is the general criticism?

I didn't follow the bogleheads thread closely, but it looked to me as if many Bogleheads who participated in the discussion missed fact that real estate delivers real returns as good as equities (that makes sense since that forum's orthodoxy centers around equity index funds) and then many Bogleheads missed realty that real estate shows little correlation with equities (which makes a GIANT difference if you try to implement MPT in your investing).

[coopdog]

Hey SeattleCPA, do you know of recommended ratios for RE/equity/fixed for MPT? I follow a MPT methodology I picked up 15 years ago and I don't recall what it said about RE.

[toganet]

Hey SeattleCPA, do you know of recommended ratios for RE/equity/fixed for MPT? I follow a MPT methodology I picked up 15 years ago and I don't recall what it said about RE.

PTF because I have been mulling the same question.  Recently decided to add rental properties to my FIRE plan, but don't know how to frame that within the overall picture.

[SeattleCPA]

I don't have any actionable insights on this (basically because I'm a corporate tax accountant and have been more than fully occupied learning the new tax law and then preparing corporate returns). But this all deserves careful consideration.

For what it's worth, two comments to move the discussion forward a millimeter or two...

1. Though I've historically used a Swensen allocation (so 30% US stocks, 15% developed markets, 15% REITS, 15% TIPS, 15% intermediate treasuries and then 10% emerging markets) for traditional asset class investments, because of direct real estate investments (including a primary residence) I've let my REIT % slip down to around 8%. That's not some well thought out, quantitative analytical result, however... just portfolio drift really. But I do think one nudges percentages around as a first step rather than making a wholesale change.

I blogged a little bit about this here: https://evergreensmallbusiness.com/asset-allocation-for-small-business-owners-e/


2. You can use Portfolio Visualizer to do monte carlo simulations of assets classes based on their historical variability and a specified mean. I've played around with that a bit (not in a way i feel comfortable talking about publicly BTW)... and I do think one can assemble a portfolio that probably dials down one's risks:

https://www.portfoliovisualizer.com/monte-carlo-simulation

BTW not to start a fight, but I'll admit that i've been applying CAPE-type thinking to come up with modest expected returns for coming decade in equities.

[the_fella]

Can I get a TL;DR?

[rab-bit]

Great link, thanks for sharing!

[beltim]

I didn't follow the bogleheads thread closely, but it looked to me as if many Bogleheads who participated in the discussion missed fact that real estate delivers real returns as good as equities (that makes sense since that forum's orthodoxy centers around equity index funds) and then many Bogleheads missed realty that real estate shows little correlation with equities (which makes a GIANT difference if you try to implement MPT in your investing).

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

[SeattleCPA]

I didn't follow the bogleheads thread closely, but it looked to me as if many Bogleheads who participated in the discussion missed fact that real estate delivers real returns as good as equities (that makes sense since that forum's orthodoxy centers around equity index funds) and then many Bogleheads missed realty that real estate shows little correlation with equities (which makes a GIANT difference if you try to implement MPT in your investing).

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

The way I read and understand the information (charts etc) on pages 20 and 21 of the pdf, I sure think so.

Seems to me that it's not much of an exaggeration to say the low correlation is almost like Holy Grail.

[toganet]

I didn't follow the bogleheads thread closely, but it looked to me as if many Bogleheads who participated in the discussion missed fact that real estate delivers real returns as good as equities (that makes sense since that forum's orthodoxy centers around equity index funds) and then many Bogleheads missed realty that real estate shows little correlation with equities (which makes a GIANT difference if you try to implement MPT in your investing).

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

The way I look at the rebalancing side of it is to consider the RE asset as the "last to move" portion of whichever category it's in.  As in, I would never go to 0% equities, so there is always some portion of that asset class whose liquidity is irrelevant.

Of course, it might be desirable to liquidate a given piece of RE in a falling market -- but that's probably not the time you should be looking to sell, anyway :) 

[SeattleCPA]

I didn't follow the bogleheads thread closely, but it looked to me as if many Bogleheads who participated in the discussion missed fact that real estate delivers real returns as good as equities (that makes sense since that forum's orthodoxy centers around equity index funds) and then many Bogleheads missed realty that real estate shows little correlation with equities (which makes a GIANT difference if you try to implement MPT in your investing).

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

The way I look at the rebalancing side of it is to consider the RE asset as the "last to move" portion of whichever category it's in.  As in, I would never go to 0% equities, so there is always some portion of that asset class whose liquidity is irrelevant.

Of course, it might be desirable to liquidate a given piece of RE in a falling market -- but that's probably not the time you should be looking to sell, anyway :)

Makes sense.

Also, the study looks at un-leveraged residential real estate... which means probably properties are cash flowing (no mortgages) so at least during accumulation phase you would be able to do some re-balancing by putting all the portfolio's income (and any additional savings) into the asset class where you were light.

[GU]

Awesome paper, thanks for posting.

[itchyfeet]

I have t read the report, but will try at some point. So for now posting to follow.

I am sure there is research out there, but I wonder how closely related growth in residential property values is to growth in household incomes. I feel sure that the increased participation and incomes of women if the workforce must have had a once off, non repeatable impact on property returns.

On this point, I have pondered (with no serious intent, and no research to back my thoughts) making a small speculative investment in a developing country where women’s workforce participation remains low but is on the rise.

[Icecreamarsenal]

Thanks for the paper; anyone proffer a TL;DR?

[DNA2RNA]

TL:DR

Unleveraged Housing has roughly same real return as stocks over the long-term with less than half the volitility,  has very little correlation with stocks, has almost no correlation with housing in other countries, and has a slight positive correlation with inflation whereas stocks and bonds have a negative coorelation.

If you try to calculate the optimal portfolio given these findings are correct, as SeattleCPA is talking about, you will find that bonds would be almost completely removed and just over half the portfolio will be diversified global real estate.

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

[SeattleCPA]

TL:DR

Unleveraged Housing has roughly same real return as stocks over the long-term with less than half the volitility,  has very little correlation with stocks, has almost no correlation with housing in other countries, and has a slight positive correlation with inflation whereas stocks and bonds have a negative coorelation.

If you try to calculate the optimal portfolio given these findings are correct, as SeattleCPA is talking about, you will find that bonds would be almost completely removed and just over half the portfolio will be diversified global real estate.

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

After I read the paper the first time, I grabbed some spreadsheets from MIT's business school and, based on numbers from the study, used those to calculate that in a two asset portfolio, i should have more than 90%-ish in real estate and less than 10% in US stocks.

If you use the higher standard deviation of a "not geographically diversified" real estate portfolio, your real estate allocation goes way down...to maybe 50% or so per my rough calculations.

But the point is, as DNA2RNA so concisely points out, you're getting same return as stocks with half the volatility and low correlation with the other stuff in your portfolio.

It's like magic pixie dust.

[boarder42]

So same return lower volatility. Anybody have a way to back test withdrawal rates with different gloabl reit index to vtsax. Bc it sounds like this would allow for 5-6% swr if the volatility is really that much lower.

[sol]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

I have not yet read the quoted paper, but I do have some thoughts on this decision

First, didn't Piketty say global real estate is still something like 80% 60% of all global wealth?  There just isn't enough intellectual property in the world to compete with the value of all buildings and lands.

Second, we've long recognized that the wealthier a person is, the more real estate they own.  Coincidence?  Consequence of better investment decisions?  The ultra-rich are basically ALL heavily invested in real estate, but this has traditionally been interpreted as a strategy to minimize volatility, not increase returns.  They don't need anywhere near 100% liquidity, but they are more fearful of 50% short term losses.

Third, I think I agree with the assertion above that RE can largely replace bonds.  Bonds have always been a philosophically questionable asset class.  Real estate is very much like a bond.  So is a pension, or an annuity.  So is your ability to work at a job (aka human capital).  There are LOTS of ways to make regular income without being dependent on the current market price of corporate shares, and bonds are perhaps the least uncorrelated option on that list because they are still dependent on the financial health of institutions.

[sol]

I will spend some time with the posted link, but as a closing thought I think it's important to note that modern corporations are not valued the same way they used to be, and that has significantly skewed the way we calculate global "wealth".  Companies like Netlix and Facebook have almost no hard assets, yet are some of the most highly valued corporations in history.  They generally don't own vast tracts of land, or have tens of thousands of employees, or buildings full of machinery and raw materials.  Their wealth is in information and organization and income stream, which skews the numbers. 

Consider that Uber is worth approximately $70 billion and owns... almost nothing.  No cars.  No employees.  No inventory.  It's literally just an app that organizes private contractors.  If it ever goes bankrupt, what assets are they going to sell to repay their creditors?  Where's the gleaming corporate HQ building, or the stockpiles of raw materials that can go up for bid?  How do you put a "book value" on Uber?

[beltim]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

And when you add liquidity and management expenses, for example in REITS, costs go up and returns go down.  It's worth noting that in order to achieve the same returns as unleveraged real estate, REITs needed 45% leverage.  So strategies like

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

are adding quite a bit of risk to get the same return.

[boarder42]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

And when you add liquidity and management expenses, for example in REITS, costs go up and returns go down.  It's worth noting that in order to achieve the same returns as unleveraged real estate, REITs needed 45% leverage.  So strategies like

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

are adding quite a bit of risk to get the same return.

but how risky is that leverage - when it is clear that there is less volatility in VNQI than there is in VTSAX.  When we talk about risk here we are primarily talking volatility - if you can eliminate some volatility and get the same return you've decreased your risk.  I understand you're equating the leverage to risk but the results speak for themselves historically- the risk is lower in RE when spread widely across an index.  On the higher expenses front we're talkin about ~4x higher expenses.  .15 vs .04 which is probably insignificant if my volatility drops low enough that i dont have real sequencing risk.  so the same return without sequencing risk - over the last 150 years at a cost of .11% seems to be a pretty fair trade off.  - again would be good if @bo_knows could put this REIT data into cFIREsim so we could back test it. b/c it should equate to a 5-6% SWR by my guestimate.

playing with portfolio charts i get a PWR of 90/10 equity to bonds of 3.8 and an swr of 4.5
If i go 50/50 with equity and REIT i get a 4.6 PWR and 5.2 SWR. 
if we go all the way to 90/10 REIT to equity i get 4.7 and 5.5

his data only goes back to sometime in the 70s though. 

it interestingly enough also tightens the window for FIRE with the number i put in it cut 2 years off of the longest time to FIRE but didnt cut the speed to FIRE once REITs were introduced.

[beltim]

  On the higher expenses front we're talkin about ~4x higher expenses.  .15 vs .04 which is probably insignificant if my volatility drops low enough that i dont have real sequencing risk.  so the same return without sequencing risk - over the last 150 years at a cost of .11% seems to be a pretty fair trade off. 

There's no way that the cost of 45% leverage is 0.11%.  That would imply an interest rate of just a quarter of one percent.

[boarder42]

  On the higher expenses front we're talkin about ~4x higher expenses.  .15 vs .04 which is probably insignificant if my volatility drops low enough that i dont have real sequencing risk.  so the same return without sequencing risk - over the last 150 years at a cost of .11% seems to be a pretty fair trade off. 

There's no way that the cost of 45% leverage is 0.11%.  That would imply an interest rate of just a quarter of one percent.

i'm talking the expense ratio of the fund. but if the fund return is on par with stock why does the leverage inside of it matter.

[beltim]

  On the higher expenses front we're talkin about ~4x higher expenses.  .15 vs .04 which is probably insignificant if my volatility drops low enough that i dont have real sequencing risk.  so the same return without sequencing risk - over the last 150 years at a cost of .11% seems to be a pretty fair trade off. 

There's no way that the cost of 45% leverage is 0.11%.  That would imply an interest rate of just a quarter of one percent.

i'm talking the expense ratio of the fund. but if the fund return is on par with stock why does the leverage inside of it matter.

It's still a cost.  A very real one.  And unlike individual mortgages, most commercial mortgages are not designed to be paid off in 30 years, and are subject to asset valuations, capital calls, and possible interest rate changes.  These are real risks, where a company may have to sell productive assets at low prices because they are over leveraged.

[Pooplips]

Post to Follow

[boarder42]

  On the higher expenses front we're talkin about ~4x higher expenses.  .15 vs .04 which is probably insignificant if my volatility drops low enough that i dont have real sequencing risk.  so the same return without sequencing risk - over the last 150 years at a cost of .11% seems to be a pretty fair trade off. 

There's no way that the cost of 45% leverage is 0.11%.  That would imply an interest rate of just a quarter of one percent.

i'm talking the expense ratio of the fund. but if the fund return is on par with stock why does the leverage inside of it matter.

It's still a cost.  A very real one.  And unlike individual mortgages, most commercial mortgages are not designed to be paid off in 30 years, and are subject to asset valuations, capital calls, and possible interest rate changes.  These are real risks, where a company may have to sell productive assets at low prices because they are over leveraged.

these are also all risks that we have with equities and leveraged companies that are probably a part of all of our portfolios - the study - someone can correct me if i'm wrong basically compared total return including costs of these different asset classes and found REIT's to be very closely related to equities return wise with respect to ROI - but with much lower volatility.

[DNA2RNA]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

And when you add liquidity and management expenses, for example in REITS, costs go up and returns go down.  It's worth noting that in order to achieve the same returns as unleveraged real estate, REITs needed 45% leverage.  So strategies like

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

are adding quite a bit of risk to get the same return.

I am rare in that I own unlevered individual real estate here in the US, which is my largest position.  VNQI is now my biggest "equity" position.

I would love to own unlevered diversified global real estate, but it is unrealistic so I will take the implicit leverage from VNQI, the other diversification and correlation benefits far outweigh that extra risk.

[beltim]

  On the higher expenses front we're talkin about ~4x higher expenses.  .15 vs .04 which is probably insignificant if my volatility drops low enough that i dont have real sequencing risk.  so the same return without sequencing risk - over the last 150 years at a cost of .11% seems to be a pretty fair trade off. 

There's no way that the cost of 45% leverage is 0.11%.  That would imply an interest rate of just a quarter of one percent.

i'm talking the expense ratio of the fund. but if the fund return is on par with stock why does the leverage inside of it matter.

It's still a cost.  A very real one.  And unlike individual mortgages, most commercial mortgages are not designed to be paid off in 30 years, and are subject to asset valuations, capital calls, and possible interest rate changes.  These are real risks, where a company may have to sell productive assets at low prices because they are over leveraged.

these are also all risks that we have with equities and leveraged companies that are probably a part of all of our portfolios - the study - someone can correct me if i'm wrong basically compared total return including costs of these different asset classes and found REIT's to be very closely related to equities return wise with respect to ROI - but with much lower volatility.

Yes, those risks are included in the equity data set.  This paper did not look at leveraged real estate, at least not in terms of volatility, and so you simply do not have any information on the long-term volatility of leveraged real estate at all, much less on REITs.  The data in the paper on REITS consists entirely of:
1) A portfolio of 15 properties in Paris, listed on the French stock exchange between 1904 and 1997; and
2) A US REIT index from ~1975 - present.
Even the authors say (page 37):

The REIT returns, however, tend to be somewhat more volatile—most likely because
they reflect changes in valuation of future earnings, as well as the current portfolio performance.
The REIT returns also seem to be affected by the general ups and downs of the stock market: for
example, the 1987 “Black Monday” crash and dot-com bust in the US, as well as the 1930s Great
Depression and 1960s stock crises in France. This suggests that the valuations of the fund’s housing
portfolios may be affected by general stock market sentiment.

Put another way, the study does not indicate the same effects for REITs, and so you cannot simply substitute REITs and expect to have the same volatility as the housing market in general.

[rab-bit]

I am rare in that I own unlevered individual real estate here in the US, which is my largest position.  VNQI is now my biggest "equity" position.

I would love to own unlevered diversified global real estate, but it is unrealistic so I will take the implicit leverage from VNQI, the other diversification and correlation benefits far outweigh that extra risk.

@DNA2RNA, I am just curious about the extent to which your real estate is diversified, either geographically, by type (e.g. residential vs. commercial), or other?

[beltim]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

And when you add liquidity and management expenses, for example in REITS, costs go up and returns go down.  It's worth noting that in order to achieve the same returns as unleveraged real estate, REITs needed 45% leverage.  So strategies like

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

are adding quite a bit of risk to get the same return.

I am rare in that I own unlevered individual real estate here in the US, which is my largest position.  VNQI is now my biggest "equity" position.

I would love to own unlevered diversified global real estate, but it is unrealistic so I will take the implicit leverage from VNQI, the other diversification and correlation benefits far outweigh that extra risk.

Fair enough.  I would point your attention to appendix G, which shows a much higher volatility for international real estate in US dollar terms than in local terms, which of course makes sense, and may (or may not) change the implications of MPT.

Additionally, see my comments to boarder42 regarding the utter inapplicability of this study to REITs.

[boarder42]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

And when you add liquidity and management expenses, for example in REITS, costs go up and returns go down.  It's worth noting that in order to achieve the same returns as unleveraged real estate, REITs needed 45% leverage.  So strategies like

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

are adding quite a bit of risk to get the same return.

I am rare in that I own unlevered individual real estate here in the US, which is my largest position.  VNQI is now my biggest "equity" position.

I would love to own unlevered diversified global real estate, but it is unrealistic so I will take the implicit leverage from VNQI, the other diversification and correlation benefits far outweigh that extra risk.

Fair enough.  I would point your attention to appendix G, which shows a much higher volatility for international real estate in US dollar terms than in local terms, which of course makes sense, and may (or may not) change the implications of MPT.

Additionally, see my comments to boarder42 regarding the utter inapplicability of this study to REITs.

Thanks @beltim if only it applied to REITs

[DNA2RNA]

I am rare in that I own unlevered individual real estate here in the US, which is my largest position.  VNQI is now my biggest "equity" position.

I would love to own unlevered diversified global real estate, but it is unrealistic so I will take the implicit leverage from VNQI, the other diversification and correlation benefits far outweigh that extra risk.

@DNA2RNA, I am just curious about the extent to which your real estate is diversified, either geographically, by type (e.g. residential vs. commercial), or other?

Not very diversified, which is maybe why I am looking at the divesification benefits the closest.  Basically I own 3 residential rental properties, 2 in a major market that I live in and know well, and another one in another major market 5 hours away

[TexasRunner]

Thanks for posting this.

I'll have a go at it later.

[brooklynguy]

I'm going to spend more time with the linked paper, but I have a few initial observations:

1.  The real estate returns examined by the paper appear to be limited to residential housing, which is only one subset of the broader category of "real estate" in general as an asset class.  The historical returns of the former may or may not be representative of historical returns of the latter (or of other subsets of the latter).

2.  The overall result that historical residential housing returns were comparable to overall stock returns is not equally true across all geographical areas examined or across all time periods examined.  As depicted in Table 5 on pg. 22 of the paper, in the United States, stocks outperformed residential housing by what I would consider a significant margin, and more so as the time periods in question became more recent (real returns on stocks vs. housing in the U.S. were:  8.39% to 6.03% across the full sample period; 8.75% to 5.62% in the post-1950 period; and 9.09% to 5.66% in the post-1980 period).

3.  Much like a bond mutual fund, the performance of a REIT is unlikely to perfectly track the performance of a directly-held portfolio of its underlying holdings, and is likely to exhibit a greater degree of positive correlation with stock returns (some of the reasons why this is the case are discussed on pages 35 and 36 of the paper, including, as beltim noted, the fact that they typically have management costs and embedded leverage that differ from a directly-held portfolio of real estate assets).

Edited to add that I see beltim already addressed point 3 in more detail as I was typing.

[DNA2RNA]

Is this really relevant for MPT? Real estate is not usually a liquid, low transaction fee asset class, and so I question whether it's really practical for even annual rebalancing.

RE traditionally offers higher returns precisely because it is illiquid.  That's a well-recognized feature of RE.  Returns go up as risk goes up and liquidity goes down.

And when you add liquidity and management expenses, for example in REITS, costs go up and returns go down.  It's worth noting that in order to achieve the same returns as unleveraged real estate, REITs needed 45% leverage.  So strategies like

When I read this paper a few months ago, I thought about it for a few days and then made VNQI (Vanguard International Real Estate Index ETF) my largest holding.

are adding quite a bit of risk to get the same return.

I am rare in that I own unlevered individual real estate here in the US, which is my largest position.  VNQI is now my biggest "equity" position.

I would love to own unlevered diversified global real estate, but it is unrealistic so I will take the implicit leverage from VNQI, the other diversification and correlation benefits far outweigh that extra risk.

Fair enough.  I would point your attention to appendix G, which shows a much higher volatility for international real estate in US dollar terms than in local terms, which of course makes sense, and may (or may not) change the implications of MPT.

Additionally, see my comments to boarder42 regarding the utter inapplicability of this study to REITs.

Utter inapplicibility is a pretty absurd statement.  The real returns in US $ terms over the 150 yr sample across 30+ countries from Appendix G look like this:

Housing: 8.11% with a 15.83% standard deviation
Stocks : 7.84% with a 25.08% standard deviation

Its hard to comprehend the implications of such numbers.  But as SeattleCPA says the optimal portfolio would be ~ 90% Real Estate/10% Stocks if those numbers hold up in the future.  So by going with roughly a ~50/50 leaves a ton of room for error, i.e. additional leverage risk, market risk from the REIT being traded in the stock market, and just about anything you can think of.

[beltim]

Additionally, see my comments to boarder42 regarding the utter inapplicability of this study to REITs.

Utter inapplicibility is a pretty absurd statement.  The real returns in US $ terms over the 150 yr sample across 30+ countries from Appendix G look like this:

Housing: 8.11% with a 15.83% standard deviation
Stocks : 7.84% with a 25.08% standard deviation

Its hard to comprehend the implications of such numbers.  But as SeattleCPA says the optimal portfolio would be ~ 90% Real Estate/10% Stocks if those numbers hold up in the future.  So by going with roughly a ~50/50 leaves a ton of room for error, i.e. additional leverage risk, market risk from the REIT being traded in the stock market, and just about anything you can think of.

I described earlier the entire basis set for the REIT comparison, and quoted the section of the study where the authors describe that REITs are more volatile, and more correlated with the stock market.  If you want to ignore the authors' warning and misuse their data, go ahead.

[DNA2RNA]

Additionally, see my comments to boarder42 regarding the utter inapplicability of this study to REITs.

Utter inapplicibility is a pretty absurd statement.  The real returns in US $ terms over the 150 yr sample across 30+ countries from Appendix G look like this:

Housing: 8.11% with a 15.83% standard deviation
Stocks : 7.84% with a 25.08% standard deviation

Its hard to comprehend the implications of such numbers.  But as SeattleCPA says the optimal portfolio would be ~ 90% Real Estate/10% Stocks if those numbers hold up in the future.  So by going with roughly a ~50/50 leaves a ton of room for error, i.e. additional leverage risk, market risk from the REIT being traded in the stock market, and just about anything you can think of.

I described earlier the entire basis set for the REIT comparison, and quoted the section of the study where the authors describe that REITs are more volatile, and more correlated with the stock market.  If you want to ignore the authors' warning and misuse their data, go ahead.

I will also quote from the same section:
"...Comparing the solid and dashed lines in Figure 12, the long-run levels of unlevered REIT and
housing returns are remarkably similar. The time trend also follows a similar pattern, especially
in France. The REIT returns, however, tend to be somewhat more volatile—most likely because
they reflect changes in valuation of future earnings, as well as the current portfolio performance.
The REIT returns also seem to be affected by the general ups and downs of the stock market: for
example, the 1987 “Black Monday” crash and dot-com bust in the US, as well as the 1930s Great
Depression and 1960s stock crises in France. This suggests that the valuations of the fund’s housing
portfolios may be affected by general stock market sentiment.
Overall, the returns on real estate investment funds serve to confirm the general housing return
level in our dataset...."

REITs have such a small sample (basically 1 country over 35 years instead of 30 countries over 150 countries) that I will rely quite a bit on the performance of the underlying real estate, you can feel free not to as most will. I understand that I am outlier.   Based on the whole section and graphs its obvious that they mean REITs will be somewhat more volatile than unlevered real estate NOT somewhat more volitile than stocks.

But if you offer your estimations of long term returns, deviations, and correlations for Stocks, REITs, and bonds that you think are better than using a 150 year history I would be happy to calculate an optimal portfolio for you.  I already estimated that returns would at least 1% less per year for housing to account for taxes and volatility would be 30% higher to account for leverage inherent in REITS.

[boarder42]

i'm gonna wait for some of the smarter people with better math analysis than myself to draw some conclusions here and present some bigger back testable data.  maybe we wont find it- you'd think if REITs were that much safer everyone would be in them instead of stocks and that would be what we all talked about around here - its not like this paper unearthed a new holy grail no one had looked at before.

[max9505672]

PTF

[Mr. Boh]

I just want to emphasize what others have said. This data is for un-leveraged residential real estate. It would be a big mistake to assume a REIT would be the same thing. REITs are much more correlated to equities. Take a look at these one year charts. The first is VNQI and the second one is VOO (S&P 500).




Do you notice any similarities (correlation)? I realize that these pictures don't prove anything, and in fact if you look at longer time frames the similarities break down, but I wouldn't bet the farm that they are not correlated. In my opinion un-leveraged residential real estate vs. REITs is apples vs. oranges.

Edit: Thanks TexasRunner for your technical help!

[TexasRunner]

They are linked from a https url.  Save them to photobucket or something similar and then relink.

[DNA2RNA]

I understand that REITs =/ unleveraged real estate.  As I said before I actually own unleveraged RE in the US, and own a insignificant amount of US REITs (in fact for most of last year I was hedging my individual holdings by shorting some US REITs).

So I understand the sentiment and was a firm believer as of last year.  The problem is I cant compare individual foreign real estate holdings with foreign REITs.  I just can't know enough about local laws, customs, transactions, etc to do it.  So its VNQI vs bonds or more equities.  And just to take Mr. Boh's comparison of VOO and VNQI, as of today via ETF.com

VNQI PORTFOLIO DATA
Weighted Average Market Cap   $11.53B
Price / Earnings Ratio                10.41
Price / Book Ratio                     1.07
Distribution Yield                       3.88%

VOO PORTFOLIO DATA
Weighted Average Market Cap   $195.93B
Price / Earnings Ratio                 25.48
Price / Book Ratio                      3.30
Distribution Yield                        1.77%

Talk about apples and oranges

[Mr. Boh]

DNA2RNA my post was not specifically directed to you. I was merely trying to show that REITs are correlated to equities. From the OP:

The other thing, really interesting to anybody who tries to employ modern portfolio theory to their investing, is that it looks like housing and equities are not very correlated.

It has been suggested that REITs could take the place of bonds in a portfolio and I was trying to point out that this is not the case.