The Realized Historical Return Is the Wrong Yardstick

[CM]

I just found a recent Project Syndicate article by Robert Shiller: https://www.project-syndicate.org/commentary/us-stock-volatility-bear-market-by-robert-j--shiller-2017-09. Shiller is famously non-committal regarding prospective returns and timing of bear markets, but like me, he is concerned enough to write words of caution on the Internet.

The header reads:

The US stock market today looks a lot like it did at the peak before all 13 previous price collapses. That doesn't mean that a bear market is imminent, but it does amount to a stark warning against complacency.

 

He provides a definition for a bear market, then examines the relationship between PE10 (CAPE) and bear markets:

This month, the CAPE ratio in the US is just above 30. That is a high ratio. Indeed, between 1881 and today, the average CAPE ratio has stood at just 16.8. Moreover, it has exceeded 30 only twice during that period: in 1929 and in 1997-2002.

But that does not mean that high CAPE ratios aren’t associated with bear markets. On the contrary, in the peak months before past bear markets, the average CAPE ratio was higher than average, at 22.1, suggesting that the CAPE does tend to rise before a bear market.

Moreover, the three times when there was a bear market with a below-average CAPE ratio were after 1916 (during World War I), 1934 (during the Great Depression), and 1946 (during the post-World War II recession). A high CAPE ratio thus implies potential vulnerability to a bear market, though it is by no means a perfect predictor.

The article also contains this data point:

According to my data, real S&P Composite stock earnings have grown 1.8% per year, on average, since 1881.

I used Siegel's 1.25% real eps growth between 1871 and 2001 in my projections above. It might be more reasonable to use Shiller's 1.8% real eps growth between 1881 and present. So, I hereby increase my annual growth projections by 0.55%, from 1.25% to 1.8%.

[CM]

CM wrote:

The base-case scenario assumes a reversion to mean valuation. If the future mean is equal to the historical mean, then the total real annual return will be slightly negative (i.e., a little worse than cumulative -17.18%, as described above). However, the future mean will probably be something a little higher than the historical mean (PE10 of 20 instead of 16.79?). If so, the total real annual S&P 500 return might be slightly positive. Incorporate 0.48% return for 10-year TIPS and the portfolio will probably return between zero and 1% real. That’s the base-case (“treading water”), not the best case.

Click to expand...

I just checked the graph of historical returns from Big ERN at https://earlyretirementnow.com/2016/12/21/the-ultimate-guide-to-safe-withdrawal-rates-part-3-equity-valuation/.

There are very few historical data points at a PE10 of 30ish; it's among the highest in the record. However, the best subsequent 10-year return was around 5% and the worst looks to be about -1.8%, with most clustering around 0-2%. So there is that.

[CM]

ko wrote:

So just to make sure I understand your bottom line here – you are saying that, at best, you expect around a 0% real gain for a 70/30 or 60/40 portfolio over the next decade?

Click to expand...

No, "at best" would be better.

I suppose the best case scenario is something unprecedented. What is the opposite of: A meteor destroys all human life?

If I recall correctly, real eps growth was a little more than 3% for the 20 years after WWII, and that's the best it has ever been for an extended period. If that occurs for the next 10 years without a fall in valuation, then the S&P 500 will provide a 5% total real return. The probability is low, but non-negligible.

So a reasonable best-case scenario would be 5% real from the S&P 500. The real return of 10-year TIPS is a given: 0.48%. Multiply by your portfolio percentages to obtain best-case portfolio return.

The base-case scenario assumes a reversion to mean valuation. If the future mean is equal to the historical mean, then the total real annual return will be slightly negative (i.e., a little worse than cumulative -17.18%, as described above). However, the future mean will probably be something a little higher than the historical mean (PE10 of 20 instead of 16.79?). If so, the total real annual S&P 500 return might be slightly positive. Incorporate 0.48% return for 10-year TIPS and the portfolio will probably return between zero and 1% real. That's the base-case ("treading water"), not the best case.

The worst-case scenario would include a bear market valuation at the 10-year horizon. If it's the 1982 valuation, then the S&P 500 will suffer a cumulative (not annual) loss of 50% or more over 10 years. The 1982 nadir was among the worst, so things are unlikely to be that bad. A more reasonable worst-case scenario would be a cumulative real S&P 500 loss of something between 20% to 50%, offset by the 0.48% annual real return on 10-year TIPS.

I think the worst-case scenario is much more likely than the best-case scenario described above. The change in valuation is the most powerful driver of the outcome over a short (e.g., 10-year) horizon. We don't know what that will be at a point in time, but in my opinion, it's likely to be substantially lower than today.

[Privatefarmer]

Do most here agree that all publicly available information is already priced into the market? If so, would you not agree that the CAPE-10 is priced into the current market? In other words, if nobody knew what the CAPE-10 was, do you think the market would be even higher than it is today? Perhaps the market has already corrected for the "high valuation" according to the CAPE? Perhaps the CAPE-10 would be 45 if we all didn't know what the CAPE-10 was.

My point is that obsessing over this stuff is detrimental to your wealth. There will ALWAYS be logical reasons why the market is over valued and logical reasons why it is under valued.

[ko]

CM wrote:

ko wrote:

So just as a point of context, what do you think real returns will be over the next 10 years for a 60/40 equity-to-bond or 70/30 portfolio?

Really just curious where you’re landing after taking all this info into consideration.

 

Click to expand…

For the S&P 500, I estimate future returns with the expected forward dividend yield plus real earnings growth plus/minus an adjustment for valuation.

The forward dividend yield is 1.98% today (S&P spreadsheet available from top listing on this page: https://www.google.com/search?q=s%26p+500+earnings+howard+silverblatt&rlz=1C1CHBF_enUS752US752&oq=S%26P+earnings+howar&aqs=chrome.1.69i57j0l2.10823j0j7&sourceid=chrome&ie=UTF-8).

Siegel reported that real eps grew 1.25%/year between 1871 and 2001 (Stocks for the Long Run, Third Edition, Table 6-1), so I usually use that figure to estimate growth.

If I recall correctly, real eps declined slightly in the 1930s and grew a little more than 3% real during the 20 years following WWII, so my best case scenario for the long-term, real S&P 500 return is about 3.23% (1.98+ 1.25) with a likely range of 2% to 5%.

That’s the best case scenario because it doesn’t account for valuation change and the valuation is very high today. (That is, this best case scenario assumes today’s valuation is the new and future mean.)

The mean PE10 from 1871-present is 16.79 (available in spreadsheet from Shiller’s website: http://www.econ.yale.edu/~shiller/data.htm), and it’s 30.73 today. Reversion to the mean tomorrow implies a 45.36% loss.

A quick spreadsheet calculation shows that reversion to the mean at a 10-year horizon implies a cumulative, real, 10-year loss of 17.18% (after including dividends and real eps growth for a decade).*

Of course, reversion to the mean is not the worst case scenario. Valuations also fall below the mean (by definition).

The PE10 nadir in 1982 was 6.62. Reversion to this bear market nadir at 10 years implies a cumulative real loss of 54.58% over the next decade (after accounting for dividends and real eps growth).

Reversion to this bear market PE10 even 29 years from now causes a 0.32% cumulative, real loss. Yes, a buy-and-hold, stay-the-course investor will lose money over a 29 year period if we happen to have a bad bear market at that horizon (assuming a historical rate of real eps growth).

—

10-year TIPS have 0.48% real yield, and 30-year TIPS provide 0.93% (https://www.bloomberg.com/markets/rates-bonds/government-bonds/us).

 

So, given a likely small annual loss on the S&P 500 and a small annual gain with bonds, the 60:40 or 70:30 portfolio is likely to tread water over the next decade–unless we see a typical bear market nadir at that horizon. In that case, the 60:40 or 70:30 portfolio will experience a significant loss.

 

*The cumulative real loss calculations don’t account for reinvestment of dividends. If dividends are reinvested, then the losses will be worse.

Click to expand...

Thanks for the thoughtful responses.

So just to make sure I understand your bottom line here - you are saying that, at best, you expect around a 0% real gain for a 70/30 or 60/40 portfolio over the next decade? I assume that's what you mean when you say "tread water"?

[Zaphod]

I dont think it has to be a derivative necessarily to get you the exposure you're wanting. There are probably absolutely normal set ups that fit the bill, though it really sounds like a structured product or some kind of barrier option, best of/worst of, but again those can be found in structured products.

The trick again is of course where is the dislocation between behavior and reality? Whats highly rated that shouldnt be? Then how to obtain exposure the cheapest and most profitable way is of course the difficult part.

 

[Dr.V. Investor]

CM wrote:

But is there a way to create a derivative that does the job from contracts that already exist in the marketplace?

Click to expand...

options, shorting s&P 500.  however, hedge fund managers are getting crushed doing this since nobody knows when the correction will happen.  other option Is put your perspective in longbets.org and say S&P 500 will not give 4% annual return in the next 10 years. See if others will be up for the challenge.  But, that is gambling. For us, being patient and sticking with the investment strategy is good enough for us now.

[Donnie]

CM wrote:

Donnie wrote:

Picking some random prior year and saying the returns are indicative of anything is not particularly helpful. 

Click to expand…

Not random. Selected for high valuation and low dividend.

It’s true that there is a substantial spread in the relationship between starting valuation and subsequent return, but the historical record is clear: There is an inverse relationship. See Figure 1.3 on page 11 of Shiller’s Irrational Exuberance (First Edition) or the graph of equity returns versus starting valuation on the first graph in this link from earlyretirementnow.com: https://earlyretirementnow.com/2016/12/21/the-ultimate-guide-to-safe-withdrawal-rates-part-3-equity-valuation/.

***

Our discussion about likely returns may have run its course, but I have a question for you (or anyone else following along). If I recall correctly, you have an advanced degree in a quantitative discipline and you work in the investment mgmt industry. If so, you are well positioned to know the answer.

I think it’s unlikely that the S&P 500 will provide a 4% total real annual return over the next 20 years. However, I think most people would be happy to take the opposite position. Therefore, a marketable derivative that pays off for a return below 4% ought to be cheap, if it exists. Does it? Is there a way for me to invest in this thesis?

 

I recognized the housing bubble in real time. This allowed me to rent and avoid losses on Chicago real estate, and it earned me kudos from the equity analysts at the office once the financial crisis occurred–but I didn’t make a dime off the insight.

Later I learned that Michael Burry and John Paulson made fortunes off the same insight by entering customized derivative contracts with investment banks. I’m not a hedge fund manager and don’t have this option. But is there a way to create a derivative that does the job from contracts that already exist in the marketplace?

Click to expand...

The problem is that prior to 1992, a high CAPE would have been anything over 20, so this type of analysis could have kept someone out of the market since 1992.  CAPE has been almost perpetually above 20 since then aside from the recession.  Right now CAPE is also somewhat misleading because of the abnormally low earnings from 2007-2009.  As these periods roll off, I would expect CAPE to normalize.

I don't know how to bet that market returns will be less than 4% over 20 years.  Option durations are typically much shorter than that.  That bet is also pretty nuanced compared to the housing bubble when house prices collapsed 20% in 3 years.

[Privatefarmer]

The one thing I haven't seen mentioned is that all this "data" is already priced into the market. Guess what? Hedge fund managers, institutional investors, people with BILLIONS of $$ to invest who actually set the market, Already know all this. And they've obviously set the market at its current valuation. Does that mean the market can't be over valued? Of course not. But I think it does mean that you and I cannot accurately and consistently predict when it's over or under valued.

Obviously we should all EXPECT stocks to out perform safe fixed income over any period of time. It will not always happen but it is what we should expect. Why? Because if we were to expect stocks to do worse than FI over any period of time then there would be no reason for us to invest in volatile stocks. So, today we should expect a risk premium. Tomorrow we should expect a risk premium. Next year and next decade we should expect a risk premium. No matter what your "calculated valuation" is, the collective market of investors will always expect a risk premium in equities. So if the alternative is putting your money in cash or FI, you should ALWAYS expect to earn more in equities. Doesn't mean you'll always be right, in fact I guarantee you will not always be right, but more often than not you should be.

So maybe dividends are "low" compared to historical rates. But inflation is low too. And interest rates are EXTREMELY LOW. so the bank down the street is paying 0.00001% on your savings, of course stocks can get away with paying a much lower dividend. Everything is relative. That does not mean a thing really, it's all smoke and mirrors. 10'years from now you can still be much wealthier than you are today if you invest in equities, because maybe inflation will continue to be at rock bottom levels and so even the small growth you're getting in stocks will be REAL growth? Who knows.

Don't waste your time trying to predict which market is over or under valued. If you think you are smarter than all the huge hedge funds and institutions that are happily investing in this "expensive" market then okay. Good for you. But I kinda doubt any of us are...

[CM]

Donnie wrote:

Picking some random prior year and saying the returns are indicative of anything is not particularly helpful.

Click to expand...

Not random. Selected for high valuation and low dividend.

It's true that there is a substantial spread in the relationship between starting valuation and subsequent return, but the historical record is clear: There is an inverse relationship. See Figure 1.3 on page 11 of Shiller's Irrational Exuberance (First Edition) or the graph of equity returns versus starting valuation on the first graph in this link from earlyretirementnow.com: https://earlyretirementnow.com/2016/12/21/the-ultimate-guide-to-safe-withdrawal-rates-part-3-equity-valuation/.

***

Our discussion about likely returns may have run its course, but I have a question for you (or anyone else following along). If I recall correctly, you have an advanced degree in a quantitative discipline and you work in the investment mgmt industry. If so, you are well positioned to know the answer.

I think it's unlikely that the S&P 500 will provide a 4% total real annual return over the next 20 years. However, I think most people would be happy to take the opposite position. Therefore, a marketable derivative that pays off for a return below 4% ought to be cheap, if it exists. Does it? Is there a way for me to invest in this thesis?

 

I recognized the housing bubble in real time. This allowed me to rent and avoid losses on Chicago real estate, and it earned me kudos from the equity analysts at the office once the financial crisis occurred--but I didn't make a dime off the insight.

Later I learned that Michael Burry and John Paulson made fortunes off the same insight by entering customized derivative contracts with investment banks. I'm not a hedge fund manager and don't have this option. But is there a way to create a derivative that does the job from contracts that already exist in the marketplace?

[Donnie]

CM wrote:

Donnie wrote:

Knowing when is the tricky part. 

Click to expand…

1999 and now.

Click to expand...

Here are some other total return numbers from years around 1999 until now using September of each year as starting point.  1998 - 4.5%.  1996 - 6.1%.  2001 - 5.5%.  Picking some random prior year and saying the returns are indicative of anything is not particularly helpful.

[CM]

Donnie wrote:

Knowing when is the tricky part.

Click to expand...

1999 and now.

[Donnie]

 

CM wrote:

If “the market” is expecting the historical 6.8% return, then it must be expecting roughly 4.8% real eps growth with no fall in valuation. That’s the math, and it implies 3.84 times the historical eps growth rate and 1.83 times the historical valuation. I’m saying that’s unreasonable to the point of absurdity.

Click to expand...

You make this point without including the relevant dividend yield.  Assuming 2x lower dividends than historical levels does not also fall into the level of "unreasonable to the point of absurdity" ?

CM wrote:

What return was the market pricing in 18 years ago when the PE10 was 44.2 and the forward dividend yield turned out to be 1.1% (16.27/1469)? According to dqydj.com, the total real return to date has been 2.9%. Meanwhile, TIPS offered a risk-free, long-term real return of 4-4.5%.

Click to expand...

Market is not perfect and cannot predict future events with certainty.  Anyone can find a time period to support any argument one wants to make. The things you are mentioning are all well known and priced into the market.  No, I do not think that market expects 4% returns.

People have been making the argument using similar logic that future returns will be lower than historical levels for a long time as I mentioned earlier.  There will likely be a time period when it turns out to be correct.  Knowing when is the tricky part.

[CM]

Donnie wrote:

Returns only come from price appreciation and dividends. If you are making the point that dividends are down so price appreciation must go up if we are to get the same return, I agree. If you are saying retaining more earnings shouldn’t result in higher earnings growth or that there isn’t an interplay between retaining earnings and EPS growth, I disagree. The market cares about total return. It is very unlikely that the market is pricing these assets to a total return of 4%. That is not an acceptable risk-adjusted return.

Click to expand...

If "the market" is expecting the historical 6.8% return, then it must be expecting roughly 4.8% real eps growth with no fall in valuation. That's the math, and it implies 3.84 times the historical eps growth rate and 1.83 times the historical valuation. I'm saying that's unreasonable to the point of absurdity.

Donnie wrote:

The market cares about total return. It is very unlikely that the market is pricing these assets to a total return of 4%. That is not an acceptable risk-adjusted return.

Click to expand...

What return was the market pricing in 18 years ago when the PE10 was 44.2 and the forward dividend yield turned out to be 1.1% (16.27/1469)?

According to dqydj.com, the total real return to date has been 2.9%. Meanwhile, TIPS offered a risk-free, long-term real return of 4-4.5%.

 

 

[Donnie]

Returns only come from price appreciation and dividends. If you are making the point that dividends are down so price appreciation must go up if we are to get the same return, I agree. If you are saying retaining more earnings shouldn't result in higher earnings growth or that there isn't an interplay between retaining earnings and EPS growth, I disagree. The market cares about total return. It is very unlikely that the market is pricing these assets to a total return of 4%. That is not an acceptable risk-adjusted return.