Current expected return of stocks

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
alex_686
Posts: 5151
Joined: Mon Feb 09, 2015 2:39 pm

Re: Current expected return of stocks

Post by alex_686 » Thu Feb 28, 2019 2:58 pm

HomerJ wrote:
Thu Feb 28, 2019 2:44 pm
What's your algorithm for determining long-term AA based on short-term expected returns?

Are you going to change your AA if the "experts" change their "expected" return forecasts? (and they will).
I will modestly push back. CAPE10 is a long term expected returns. Forward expectations of risk and return by asset class does evolve over time. I would not encourage any radical shifts in one's AA, but the optimal AA does evolve over time.

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 3:44 pm

HomerJ wrote:
Thu Feb 28, 2019 2:44 pm
305pelusa wrote:
Thu Feb 28, 2019 2:22 pm
dbr wrote:
Thu Feb 28, 2019 10:40 am
What has to be predicted is a probability distribution of returns. Doing that requires postulating the form of the distribution and estimates of the parameters of the distribution (expected return, expected variability of return) The estimates of the parameters are subject to estimation error. As an example 5% and 4% for expected return are probably estimates that are within estimation error of each other.

Actual results will vary from the average expected due to being a random sample from a distribution and due to the estimates of what that distribution is being themselves uncertain. So the question is what kind of predictions are being attempted from this data?
I will assume a normal distribution, will take the expected forecast returns and standard deviation (which is fairly constant throughout history). From there, I can figure out what I think is a reasonable AA.
What's your algorithm for determining long-term AA based on short-term expected returns?

Are you going to change your AA if the "experts" change their "expected" return forecasts? (and they will).
I don't have one. My algorithm/thought process determines long term AA from long term expected returns. That's why I opened a thread asking about forecasted long term returns.

Thesaints
Posts: 2921
Joined: Tue Jun 20, 2017 12:25 am

Re: Current expected return of stocks

Post by Thesaints » Thu Feb 28, 2019 3:57 pm

305pelusa wrote:
Thu Feb 28, 2019 3:44 pm
My algorithm/thought process determines long term AA from long term expected returns.
As such you should calculate your AA on a range of possible long term returns and assess the differences you get with each calculation. It is possible that your AA's will be quite similar, or otherwise you will have to use other criteria (safety, possible upside, etc.) to pick the most appropriate for you.

Keep in mind that a 1-year estimate is practically useless, being between -30% and +30% almost every single year (see, another case where expected value does not help at all), but a 30-year estimate is also extremely imprecise. The chance of an absolutely surprising event happening growing with longer time periods. 10 years are kind of a soft spot. Long enough time to damp the single year volatility, but a short enough time that the assumption of the financial environment not undergoing dramatic changes doesn't sound totally naive.

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 4:04 pm

Thesaints wrote:
Thu Feb 28, 2019 2:40 pm

Your estimate is actually rather optimistic.
Huh? I haven't even told you how I'd figure it out. You don't know if I purposely double the current Standard Deviation, for extra precaution (which effectively fattens the tails). You don't know if I multiply the expected return by 0.8 to make it even more conservative.

If you care to know, the process I used is telling me to be approximately in 40% stocks based on returns mentioned by other posters. Do you have less than that? If so, I'll eat my words.

Thesaints
Posts: 2921
Joined: Tue Jun 20, 2017 12:25 am

Re: Current expected return of stocks

Post by Thesaints » Thu Feb 28, 2019 4:40 pm

305pelusa wrote:
Thu Feb 28, 2019 4:04 pm
Thesaints wrote:
Thu Feb 28, 2019 2:40 pm

Your estimate is actually rather optimistic.
Huh? I haven't even told you how I'd figure it out. You don't know if I purposely double the current Standard Deviation, for extra precaution (which effectively fattens the tails). You don't know if I multiply the expected return by 0.8 to make it even more conservative.

If you care to know, the process I used is telling me to be approximately in 40% stocks based on returns mentioned by other posters. Do you have less than that? If so, I'll eat my words.
The 2008 event had been estimated to be 6 sigma away, if I remember correctly. The normal distribution is essentially predicated on the lack of statistical dependence of the contributing stochastic processes. Financial markets are anything but. Their sub-processes can be tightly correlated and in fact their correlation tends to be stronger in extreme situations.
If unemployment stays around the forecasted value its effect on stock prices is small. To a point that for small variations it is not even possible to say if higher unemployment helps or threatens valuations. But if unemployment skyrockets then you can be sure that its correlation with plunging stock prices becomes close to perfect.

https://sixfigureinvesting.com/2016/07/ ... ck-swans/
The link above is quite helpful; it shows that rare financial events are a lot (many many many many times) more common than expected, when one uses the wrong distribution.

I don't know if 40% is too little or too much. What I can tell you is that by using 4.67% you have eliminated the possibility of stocks returning -2%. What happens then ? What's your margin of safety ?

User avatar
HomerJ
Posts: 13560
Joined: Fri Jun 06, 2008 12:50 pm

Re: Current expected return of stocks

Post by HomerJ » Thu Feb 28, 2019 5:34 pm

305pelusa wrote:
Thu Feb 28, 2019 3:44 pm
HomerJ wrote:
Thu Feb 28, 2019 2:44 pm
305pelusa wrote:
Thu Feb 28, 2019 2:22 pm
dbr wrote:
Thu Feb 28, 2019 10:40 am
What has to be predicted is a probability distribution of returns. Doing that requires postulating the form of the distribution and estimates of the parameters of the distribution (expected return, expected variability of return) The estimates of the parameters are subject to estimation error. As an example 5% and 4% for expected return are probably estimates that are within estimation error of each other.

Actual results will vary from the average expected due to being a random sample from a distribution and due to the estimates of what that distribution is being themselves uncertain. So the question is what kind of predictions are being attempted from this data?
I will assume a normal distribution, will take the expected forecast returns and standard deviation (which is fairly constant throughout history). From there, I can figure out what I think is a reasonable AA.
What's your algorithm for determining long-term AA based on short-term expected returns?

Are you going to change your AA if the "experts" change their "expected" return forecasts? (and they will).
I don't have one. My algorithm/thought process determines long term AA from long term expected returns. That's why I opened a thread asking about forecasted long term returns.
Ah I understand. However, no one knows what long-term returns will be. Certainly not to the level of precision that you require for your algorithm to be useful.

But if you do decide to believe certain experts, what do you do when their prediction changes in 5 years after a crash? Do you change your "long-term" AA in 5 years?

That's a serious question. Even if CAPE10 is mildly predictive (not precise enough to give you a good enough answer, but let's ignore that), CAPE10 will change in 5 years if there is a crash.

So why would you stick with a "long-term" AA that you picked back in 2019 based on different numbers?

And doesn't it bug you that long-term expected returns can change over the short-term?
The J stands for Jay

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 6:58 pm

Thesaints wrote:
Thu Feb 28, 2019 4:40 pm
305pelusa wrote:
Thu Feb 28, 2019 4:04 pm
Thesaints wrote:
Thu Feb 28, 2019 2:40 pm

Your estimate is actually rather optimistic.
Huh? I haven't even told you how I'd figure it out. You don't know if I purposely double the current Standard Deviation, for extra precaution (which effectively fattens the tails). You don't know if I multiply the expected return by 0.8 to make it even more conservative.

If you care to know, the process I used is telling me to be approximately in 40% stocks based on returns mentioned by other posters. Do you have less than that? If so, I'll eat my words.
The 2008 event had been estimated to be 6 sigma away, if I remember correctly. The normal distribution is essentially predicated on the lack of statistical dependence of the contributing stochastic processes. Financial markets are anything but. Their sub-processes can be tightly correlated and in fact their correlation tends to be stronger in extreme situations.
If unemployment stays around the forecasted value its effect on stock prices is small. To a point that for small variations it is not even possible to say if higher unemployment helps or threatens valuations. But if unemployment skyrockets then you can be sure that its correlation with plunging stock prices becomes close to perfect.

https://sixfigureinvesting.com/2016/07/ ... ck-swans/
The link above is quite helpful; it shows that rare financial events are a lot (many many many many times) more common than expected, when one uses the wrong distribution.

I don't know if 40% is too little or too much. What I can tell you is that by using 4.67% you have eliminated the possibility of stocks returning -2%. What happens then ? What's your margin of safety ?
Back in 2008, the historical returns were, let's say ~7% (historical-based). The st. dev (VIX) was ~12in the 2006 era. Double it as my "dirty-quick" way of fattening the tails. What's the probability of 2008 happening under my model?

It's 5%

There has only been two crashes in the past ~100 years of that magnitude. That's a 2% chance, historically.

Upon second thought, thank you for bringing up the point about 2008. Clearly my model is being way too pessimistic and conservative.

Also just fyi, if the VIX was ~12% and the returns ~7%, 2008 is around 3 sigmas away. That's a 0.15% chance. Pretty unlikely but not absurd either.

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 7:16 pm

HomerJ wrote:
Thu Feb 28, 2019 5:34 pm

But if you do decide to believe certain experts, what do you do when their prediction changes in 5 years after a crash? Do you change your "long-term" AA in 5 years?

That's a serious question. Even if CAPE10 is mildly predictive (not precise enough to give you a good enough answer, but let's ignore that), CAPE10 will change in 5 years if there is a crash.

So why would you stick with a "long-term" AA that you picked back in 2019 based on different numbers?

And doesn't it bug you that long-term expected returns can change over the short-term?
These are excellent questions.

The reality is that at some point, I will make an arbitrary AA decision (say that's 60% stocks) based on forecasts and historical returns of the current year. From the on, I will stick to it. Every so often (a few years), I might reconsider modifying by +-X% if forecasts have changed a lot.

But to be frank, I doubt that will occur much because I do value historical returns as well and those, obviously, don't change that much over the short term. So I actually expect my AA to change more due to lifestyle factors (decreased savings rate, mortgage, longer retirement, etc) than from this process.

It is paradoxical that I would stick to a "long-term AA" picked in a previous year, with different numbers. But I have to make an arbitrary decision at some point so I'm not too bothered.

Once again, I take multiple expected returns in mind. Some, like historical returns or estimates from professionals (like Bogle) obviously won't change much in the short-term. Others, like CAPE10, change significantly. It's not an exact science; nothing about this process is. It's simply some guidance.

If I know the historical returns have been 7%, VGD says 3-5%, Bogle says 4%, earnings say 4.6%, dividend-model says 2.2%... Then that gives me pause that perhaps I shouldn't just take the historical one for granted and maybe I want to adjust my expectations down a bit.

Again, not an exact science at all.

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 7:20 pm

tibbitts wrote:
Thu Feb 28, 2019 8:39 am
305pelusa wrote:
Wed Feb 27, 2019 3:23 pm
Hello,
Could anyone point me to good resources for estimating a crude approximation of the forward-looking, long term equity premium?

I get that no one knows for certain. I just want to know a reasonable ballpark number to help guide my Asset Allocation decision a bit more.

Thank you
One issue is whether "long term" is longer than your remaining investing lifetime (I'd say so, no matter your age), and if so, are you really investing for the long term or for some short or intermediate term?
I'm 24 so (fingers crossed) hoping to invest in this game for the next 60+ years. I'm personally comfortable enough calling that a "long term" investment horizon. But you make a great point.

Thesaints
Posts: 2921
Joined: Tue Jun 20, 2017 12:25 am

Re: Current expected return of stocks

Post by Thesaints » Thu Feb 28, 2019 7:23 pm

305pelusa wrote:
Thu Feb 28, 2019 6:58 pm
Back in 2008, the historical returns were, let's say ~7% (historical-based). The st. dev (VIX) was ~12in the 2006 era. Double it as my "dirty-quick" way of fattening the tails. What's the probability of 2008 happening under my model?

It's 5%

There has only been two crashes in the past ~100 years of that magnitude. That's a 2% chance, historically.

Upon second thought, thank you for bringing up the point about 2008. Clearly my model is being way too pessimistic and conservative.

Also just fyi, if the VIX was ~12% and the returns ~7%, 2008 is around 3 sigmas away. That's a 0.15% chance. Pretty unlikely but not absurd either.
I'm sorry, can't understand why VIX is at all relevant. It does not give any magical insight in future volatility (the one that matters for you), but it is simply an average of implied volatilities on certain option contracts.

Also, your numbers are not too clear:

- historical annual return in 2008 was about the same as now, that is ~10%.
- from hi (may 2007) to low (march 2009) the S&P lost 50%, for an annualized return of -33%.
True, I'm neglecting dividends, but how does twice a 12% volatility on a 10% expected return give you a -33% ?

- Finally, you say that your model contains a 5% chance of a 50% loss ? Over which time span ? And you still get a final 4% annualized average ?
The fact that large drops only happened twice in 100 years (what about the .com ?) doesn't help much. There has been no ruinous earthquake in the bay area in the last 112 years. I still carry insurance though.

refinedchain
Posts: 7
Joined: Tue Feb 17, 2015 10:32 am

Re: Current expected return of stocks

Post by refinedchain » Thu Feb 28, 2019 7:26 pm

If you just want numbers and like fancy sharpe plots, I like JP Morgan’s forecasts, see Exhibit 6:

https://am.jpmorgan.com/us/institutiona ... ve-summary

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 8:00 pm

Thesaints wrote:
Thu Feb 28, 2019 7:23 pm

I'm sorry, can't understand why VIX is at all relevant. It does not give any magical insight in future volatility (the one that matters for you), but it is simply an average of implied volatilities on certain option contracts.
It's just one choice I made based on an author I've read. The actual historical standard deviation is much higher (~20%). We can proceed with that if you'd like. It'll simply make 2008 "more likely".
Thesaints wrote:
Thu Feb 28, 2019 7:23 pm
- historical annual return in 2008 was about the same as now, that is ~10%.
Another poster said 7% was the historical real return since the 50s. I didn't care to double check. That's my bad. Good catch.
Thesaints wrote:
Thu Feb 28, 2019 7:23 pm
- from hi (may 2007) to low (march 2009) the S&P lost 50%, for an annualized return of -33%.
True, I'm neglecting dividends, but how does twice a 12% volatility on a 10% expected return give you a -33% ?
Well clearly a 2 year loss of 50% doesn't annualize to 33% so not sure where you're getting this.
In that 2 year period, the S&P lost 40%, for an annualized loss of 22%. If you take dividend reinvestment into account, it's more like a 37% loss over the two years, annualized to 20%. I'm using this:
https://dqydj.com/sp-500-return-calculator/

Let's see how my model predicts 2008. Expected return is 10%. St Dev is 20%. I double the St. Dev to fatten the tails to 40%. What's the probability of a -37% event occurring with those parameters?

It's 12%

My model is obviously very pessimistic. It's not optimistic, like you said before.

Heck, even if you don't double the St. Dev. and just try to predict 2008 based on historical returns and historical st. dev., using a normal distribution (what you keep saying is blasphemy), you find that there's a 1% chance. That's pretty reasonable. It's about 2.5 st. dev. away from the mean.

So I take back what I said. I really like NOT fattening the tails by doubling the st. dev. and just using expected returns and standard deviation, without any changes to make the model more conservative/pessimistic.

I appreciate your time and questions.

Thesaints
Posts: 2921
Joined: Tue Jun 20, 2017 12:25 am

Re: Current expected return of stocks

Post by Thesaints » Thu Feb 28, 2019 8:24 pm

305pelusa wrote:
Thu Feb 28, 2019 8:00 pm
It's just one choice I made based on an author I've read. The actual historical standard deviation is much higher (~20%).
You mean annual volatility ? That "historical" 20% is just another average. You care about extremes.

Another poster said 7% was the historic ... ood catch.
Maybe it is the real return rate.
Well clearly a 2 year loss of 50% doesn't annualize to 33% so not sure where you're getting this.
In that 2 year period, the S&P lost 40%, for an annualized loss of 22%. If you take dividend reinvestment into account, it's more like a 37% loss over the two years, annualized to 20%. I'm using this:
https://dqydj.com/sp-500-return-calculator/
S&P went from 1510 to 757 in 21 months. That's -33% annualized. In fact the peak was later in October '17 at 1540. That would make it a 51% drop over 17 months, i.e. -39% annualized.
Let's see how my model predicts 2008. Expected return is 10%. St Dev is 20%. I double the St. Dev to fatten the tails to 40%. What's the probability of a -37% event occurring with those parameters?
What does it mean ?? The tails are the distribution tails, the possible outcomes you don't know which one of them will materialize. Those tails are fat by themselves, it is not something you make them do. Depending on the distribution you adopt it might take many sigmas to include them in your analysis and then you still have to assign a probability to them, which is not the probability according to a gaussian curve.

User avatar
Topic Author
305pelusa
Posts: 1266
Joined: Fri Nov 16, 2018 10:20 pm

Re: Current expected return of stocks

Post by 305pelusa » Thu Feb 28, 2019 9:32 pm

Thesaints wrote:
Thu Feb 28, 2019 8:24 pm
305pelusa wrote:
Thu Feb 28, 2019 8:00 pm
It's just one choice I made based on an author I've read. The actual historical standard deviation is much higher (~20%).
You mean annual volatility ? That "historical" 20% is just another average. You care about extremes.

Another poster said 7% was the historic ... ood catch.
Maybe it is the real return rate.
Well clearly a 2 year loss of 50% doesn't annualize to 33% so not sure where you're getting this.
In that 2 year period, the S&P lost 40%, for an annualized loss of 22%. If you take dividend reinvestment into account, it's more like a 37% loss over the two years, annualized to 20%. I'm using this:
https://dqydj.com/sp-500-return-calculator/
S&P went from 1510 to 757 in 21 months. That's -33% annualized. In fact the peak was later in October '17 at 1540. That would make it a 51% drop over 17 months, i.e. -39% annualized.
Let's see how my model predicts 2008. Expected return is 10%. St Dev is 20%. I double the St. Dev to fatten the tails to 40%. What's the probability of a -37% event occurring with those parameters?
What does it mean ?? The tails are the distribution tails, the possible outcomes you don't know which one of them will materialize. Those tails are fat by themselves, it is not something you make them do. Depending on the distribution you adopt it might take many sigmas to include them in your analysis and then you still have to assign a probability to them, which is not the probability according to a gaussian curve.
You assert that simply using mean and St. Dev, with an assumption that the distribution is normal, is not wise. The distribution has fatter tails than normal ones do. I concur.

Your recommendation is to instead come up with the true distribution of values, accounting for fat tails effectively. That this distribution would be much more accurate and hence, now, mean and St. Dev can be used with this fat-tailed model model for better predictions.

My cheap man's alternative is that I take whatever St. Dev you were going to apply with your fat-tailed, representative model, I double it, and then re-compute the distribution assuming it's normal. Since the mean is the same, doing so effectively spreads out the likelihood of events far from the mean. And voila, I've created a distribution model that approximates the not-Gaussian true distribution of stocks. I have overweighted results far from the mean and underweighted results near the mean.

Doubling is not a magic number. Clearly I've overestimated the probability of the fat tails. So somewhere between 1 and 2. Since none of this has to be very exact, I like the approach as a simple guesstimate.

Put into simpler words. I take the expected returns and volatility and assume it's normal. This would underestimate the true risks, because of the fat tails. So I make it up by overestimating the volatility, i.e. assuming the volatility will be much larger than historical models would say

balbrec2
Posts: 225
Joined: Mon Nov 13, 2017 3:03 pm

Re: Current expected return of stocks

Post by balbrec2 » Fri Mar 01, 2019 2:04 pm

This is interesting to say the least. Take it for what its worth. Not sure how actionable it is!
https://www.gurufocus.com/stock-market-valuations.php

Thesaints
Posts: 2921
Joined: Tue Jun 20, 2017 12:25 am

Re: Current expected return of stocks

Post by Thesaints » Fri Mar 01, 2019 4:53 pm

305pelusa wrote:
Thu Feb 28, 2019 9:32 pm
Put into simpler words. I take the expected returns and volatility and assume it's normal. This would underestimate the true risks, because of the fat tails. So I make it up by overestimating the volatility, i.e. assuming the volatility will be much larger than historical models would say
The problem is that a gaussian, however spread wide, is always taller in the middle and becomes lower the more you move away.
Fat tails are actually more probable than a range of intermediate results.
It's kind of when you skydive: either you land perfectly safe, or at most twist an ankle, or you die. There is no chance of getting a collapsed lung, or a perforated ulcer.

Counterpoint
Posts: 54
Joined: Sun Jun 05, 2016 1:42 pm

Re: Current expected return of stocks

Post by Counterpoint » Fri Mar 01, 2019 10:04 pm

One piece for the OP to check out: The Credit Suisse 2019 Global Investment Returns Yearbook, published annually together with Professors Dimson, Marsh and Staunton. It provides a good historical perspective on the equity premium in various countries/regions, including the US, over the 1900-2018 period.
You need to download the summary section from https://www.credit-suisse.com/corporate ... 01902.html

Professors DMS also have the following advice on pg 5 of the Summary: “The working premise that the authors still believe investors should factor into their long-term thinking and modelling is an annualized equity premium relative to cash of around 3 1⁄2%. This is a consistent view they have held throughout this millennium and has more or less proven to be the case. If this is disappointing based on recent history, it still points to equities historically doubling relative to cash over 20 years.”

Another piece worth reading is this recent article from Christine Benz at Morningstar (as is seeing the large variation in long-term market forecasts): https://www.morningstar.com/articles/90 ... rns-2.html

Quoted from the above article:
“It's certainly a mistake to try to predict the market in an effort to determine whether, when, and how much to hold in stocks and other asset classes. Even professional investors have struggled with tactical asset allocation, casting doubt on the ability of individual investors or even financial advisors to outperform strategic asset allocation with the approach. 

But the fact is, even long-term, strategically minded investors need some type of market-return forecast to craft a financial plan. Without any view on how much stocks, bonds, and cash are apt to return, it's impossible to know how much you'll need to save and for how long….”

I see Benz’s point that long-term forecasts of returns are probably more useful for financial planning. Nevertheless, I understand why you may want to assess various estimates of the expected value of the equity premium, in order to feel more comfort with your long-term stock allocation %.

I would however be extremely wary of using prevailing market valuations as inputs to a 40-year asset allocation path. You may end up performance-chasing - but you probably know this already. You seem aware that the key to establish an appropriate long-term asset allocation depends on your personal situation: Time horizon of accumulation and decumulation, risk tolerance, job stability, other financial circumstances and responsibilities, etc. Also, try to imagine how you would react if the stock market kept going substantially up, not just down: Would you stick with your AA? Anticipating your regret is a big factor in sticking with your plans.

Given your quantitative strengths, the most important variables in your long-term financial success could well be behavioral, especially your investment management process. So I’d suggest spending at least as much time in reading about and absorbing these issues, if you haven’t already. As a former Wall Street quant, I recognize the power of financial modeling, but I’m also very aware of its seduction and limitations.

Good luck.

User avatar
grayfox
Posts: 5261
Joined: Sat Sep 15, 2007 4:30 am

Re: Current expected return of stocks

Post by grayfox » Sat Mar 02, 2019 11:04 am

I came up with my own model for 10-Year Real Return of S&P500. It's a simple Discounted Cash Flow model. The last time I looked at this was back in 2016, so thanks for bringing this topic up in 2019. It gives me a chance to re-visit this model.

The model is really more of a What-If than a forecasting model. The model inputs are 1) the growth rate of real earnings over the next 10 years and 2) the change in CAPE.

According to multpl.com, CAPE is currently 30.64. If I input real earnings growth = 1.63%, which is I think about historical average, here is what the model shows for the 10-year real returns (in 2029) for CAPE remaining unchanged, increasing to 40 or decreasing to 20.

Code: Select all

Real Earnings Growth Rate = 1.63% p.a. (average)

CAPE.2029	20	30.64	40
10Y Real CAGR 	0.03	3.29	5.74
For average growth rate, if CAPE stays around 30, the real return is about half historic average real return of 6.7%
If CAPE goes to down to 20, you will have zero real return.

:arrow: The only way to get returns similar to historical average is for valuations to keep expanding. Or have exceptional earnings growth.

:idea: BTW, note that the commonly used forecast E.R. = 1/CAPE implies average earnings growth rate and CAPE, i.e. valuations, remaining unchanged.

User avatar
grayfox
Posts: 5261
Joined: Sat Sep 15, 2007 4:30 am

Re: Current expected return of stocks

Post by grayfox » Sat Mar 02, 2019 11:40 am

bgf wrote:
Wed Feb 27, 2019 7:32 pm
the 2018 semper augustus client letter does a good job of breaking down and explaining the historical ROE (return on equity) of the sp500. it also goes into great detail as to the downward adjustments that should be made to current publicized sp500 earnings.

if you buy the sp500 at multiple "x" and assume you can sell it later down the road at the same multiple, over a long time period you will earn the ROE, minus certain drags like writedowns, reinvested dividends, and share buybacks performed at a premium to book value.

if you arrive at an adjusted earnings number that you feel comfortable with and find the ROE, then you can calculate the expected return based on the premium you paid for the index above book value.

for example, if the ROE is 15% and you pay 3x book value, your expected return is 5%. over a long time, your return will approach the 15% number, minus the drags.

historically, sp500 ROE has been a stable 13-14%, according to the client letter. the drags however, have recently been quite large... also, sp500 price to book value is currently 3.3x. YIKES
I came up with nearly the same thing. No matter what Price/Book you pay, in the long run the return on your investment will approach the ROE. (I did not take into account the drags you mention.)

Vanguard shows S&P500 Index had ROE = 16.6%. So why can't we get 16.6%?

Obviously if you had paid exactly 1X for the Equity, you would have gotten the ROE.
But if P/B = 2, then you paid 2x for the Equity, You can think if it as paying a 100% premium.
Then amortizing that 100% over the next N years that you hold the investment.
In that case, your return will only asymptotically approach ROE over the long run.

The only fly in the ointment is, that when you paid 2x or 3x Book Value, the long run N was hundreds of years. In the short run, like 10 or 20 years or even your whole puny lifetime, the return was a lot less than ROE.

Try it in a spreadsheet and you will see how long it takes to approach ROE for various P/B.

User avatar
willthrill81
Posts: 15090
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: Current expected return of stocks

Post by willthrill81 » Sat Mar 02, 2019 11:48 am

alex_686 wrote:
Wed Feb 27, 2019 10:17 pm
I will suggest Shiller's Cape 10. It has a has a predictive power of 60%, which I find amazing for such simple and easily understood model.
It would be more accurate to say that it explains about 40% of the variance in historic subsequent ten year periods. We cannot discount the fact that Shiller data-mined CAPE, so of course it explained significant variance in the past or else we wouldn't be talking about it. That being said, it's held up reasonably well since Shiller proposed it, although it's been very far from perfect. For instance, it's been above its historic average in the U.S. almost continuously since 1992, yet real returns since then have been slightly above average. That doesn't mean that CAPE is worthless, but it's far from precise.
alex_686 wrote:
Wed Feb 27, 2019 10:17 pm
Ichiller's model is predicting 3 %to 4% real returns over the next 10 years.
1/CAPE is currently 3.26%.

A metric that has historically had far greater predictive power than CAPE (explaining over 80% of the variance in subsequent 10 year returns) is the average investor allocation to stocks. It's currently predicting 3.08% returns over the next decade.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

bgf
Posts: 1170
Joined: Fri Nov 10, 2017 9:35 am

Re: Current expected return of stocks

Post by bgf » Sat Mar 02, 2019 11:52 am

grayfox wrote:
Sat Mar 02, 2019 11:40 am
bgf wrote:
Wed Feb 27, 2019 7:32 pm
the 2018 semper augustus client letter does a good job of breaking down and explaining the historical ROE (return on equity) of the sp500. it also goes into great detail as to the downward adjustments that should be made to current publicized sp500 earnings.

if you buy the sp500 at multiple "x" and assume you can sell it later down the road at the same multiple, over a long time period you will earn the ROE, minus certain drags like writedowns, reinvested dividends, and share buybacks performed at a premium to book value.

if you arrive at an adjusted earnings number that you feel comfortable with and find the ROE, then you can calculate the expected return based on the premium you paid for the index above book value.

for example, if the ROE is 15% and you pay 3x book value, your expected return is 5%. over a long time, your return will approach the 15% number, minus the drags.

historically, sp500 ROE has been a stable 13-14%, according to the client letter. the drags however, have recently been quite large... also, sp500 price to book value is currently 3.3x. YIKES
I came up with nearly the same thing. No matter what Price/Book you pay, in the long run the return on your investment will approach the ROE. (I did not take into account the drags you mention.)

Vanguard shows S&P500 Index had ROE = 16.6%. So why can't we get 16.6%?

Obviously if you had paid exactly 1X for the Equity, you would have gotten the ROE.
But if P/B = 2, then you paid 2x for the Equity, You can think if it as paying a 100% premium.
Then amortizing that 100% over the next N years that you hold the investment.
In that case, your return will only asymptotically approach ROE over the long run.

The only fly in the ointment is, that when you paid 2x or 3x Book Value, the long run N was hundreds of years. In the short run, like 10 or 20 years or even your whole puny lifetime, the return was a lot less than ROE.

Try it in a spreadsheet and you will see how long it takes to approach ROE for various P/B.
the letter explains why investors don't actually get the ROE. there are several, but the easiest to explain are buybacks and dividends. your initial purchase in the past might approach the ROE, but every buyback or dividend reinvested at premium to book value sets you back. also, writeoffs of assets reduce ROE as well.
“TE OCCIDERE POSSUNT SED TE EDERE NON POSSUNT NEFAS EST"

User avatar
grayfox
Posts: 5261
Joined: Sat Sep 15, 2007 4:30 am

Re: Current expected return of stocks

Post by grayfox » Sat Mar 02, 2019 3:18 pm

bgf wrote:
Sat Mar 02, 2019 11:52 am

the letter explains why investors don't actually get the ROE. there are several, but the easiest to explain are buybacks and dividends. your initial purchase in the past might approach the ROE, but every buyback or dividend reinvested at premium to book value sets you back. also, writeoffs of assets reduce ROE as well.
Thanks for pointing that out. Those guys sound like they know what they're talking about. I was not aware of those sources for the drag on the ROE. That clears up a big mystery that I wondered about.

For others, Semper Augustus letter to clients, Feb-2018.

Post Reply