Post New

I was looking at a graph of S&P 500 dividend yields recently.
Obviously earnings go up and down a lot, so they need a lot of smoothing to make sense of them, and it gets to the point that the smoothing involves too many decisions.
Dividends go up and down as well, though not as much.
(sales are even less cyclical, but ignore that for the moment).

The thing that struck me about the dividend yield graph is that it has been so flat in recent years, at least compared to earnings or dividends themselves.
Maybe managers pick a dividend yield, then shareholders have been (implicitly) bidding prices to a level that keeps the yield on the current price in a narrow-ish range. When it gets outside that "normal" range, it seems that prices adjust back towards it. In the last 10 years, the average yield has been 1.87%, so that's the general range the market seems to think is normal. It was amazingly flat at around 2.0% from around 2009-Q3 till the pandemic (prices fell, so yields spiked). The credit crunch caused another spike, but before that it was very steady around 1.8-1.9% from about 2002-Q3 to end 2007, not so different.

The result is that a number as amazingly simple as the current reported S&P 500 dividend yield is a not-bad predictor of forward market returns. Or at least it has been in the last 1-3 decades, anyway. The forward time horizon I looked at was 1.5 years: the average of the annualized rates of real total return for all hold periods of 1-2 years. So, for example, the likely rate of return you'd get now selling at a random date sometime between Feb 2024 and Feb 2025...the middle of which is 18 months from now.

Here is a graph of the observed return results, starting from three different dates in the past.
http://stonewellfunds.com/DivYieldAndForwardReturn...
X axis is starting reported dividend yield, Y axis is average of annualized 18-month forward real total returns starting from that initial dividend yield.

I did a fit through each of those lines, and averaged the three fits to create this graph, suitable for pinning to your wall.
http://stonewellfunds.com/DivYieldAndForwardReturn...
Given the dividend yield on a given day what is the likely 18-month forward real total return for SPY?
The red point on the Y axis is located at today's dividend yield, though the chart could be used any time.
Using the latest figure (1.45% at the moment from Pinnacle, who uses Dow Jones News I believe), the forecast for the next ~1.5 years is an annualized rate of real total return of -3.28%/year.
It appears that this very simple model thinks the current dividend yield is low enough that there's likely a bear market coming. We'll see what next year has to offer.

As an aside:
Since dividends do go up and down somewhat with the business cycle, I figured that a smoothing of the real dividends would let me create a cyclically adjusted dividend yield (it does) which would make a better predictor (it doesn't). It seems the market reacts to the *current* level of dividends, not the "supportable" level of dividends. The side effect is that this particular model is dead easy: just look up the current dividend yield of the S&P 500 and look it up on the second chart above. (trailing, not forward).

As always, the prediction will be wrong, but the notion is this: given no other information, it's a 50/50 chance whether the market return will be higher or lower than the number you get.

Jim

Interesting study Jim. I confess, I was highly skeptical. But examining decile studies of SP500 forward total return CAGRs from various years to August 2022 confirms the overall conclusion.

2010                     
From  To   From   To     CAGR
0     0.1  1.31   1.526  -1.68%
0.1   0.2  1.526  1.83   12.24%
0.2   0.3  1.83   1.91   12.19%
0.3   0.4  1.91   1.97   13.76%
0.4   0.5  1.97   2.01   10.13%
0.5   0.6  2.01   2.05   10.84%
0.6   0.7  2.05   2.1    12.43%
0.7   0.8  2.1    2.16   16.84%
0.8   0.9  2.16   2.23   19.38%
0.9   1    2.23   2.67   24.07%
                         
2000                     
From  To   From   To     CAGR
0     0.1  1.07   1.37   -12.81%
0.1   0.2  1.37   1.664  1.33%
0.2   0.3  1.664  1.78   11.79%
0.3   0.4  1.78   1.85   5.27%
0.4   0.5  1.85   1.915  8.99%
0.5   0.6  1.915  1.99   9.34%
0.6   0.7  1.99   2.05   9.36%
0.7   0.8  2.05   2.12   11.65%
0.8   0.9  2.12   2.23   15.80%
0.9   1    2.23   3.58   20.77%
                         
1990                     
From  To   From   To     CAGR
0     0.1  1.07   1.369  -8.11%
0.1   0.2  1.369  1.64   7.25%
0.2   0.3  1.64   1.8    12.66%
0.3   0.4  1.8    1.89   7.72%
0.4   0.5  1.89   1.98   9.77%
0.5   0.6  1.98   2.06   11.14%
0.6   0.7  2.06   2.18   15.41%
0.7   0.8  2.18   2.41   20.82%
0.8   0.9  2.41   2.931  18.94%
0.9   1    2.931  4.09   14.88%
                         
1980                     
From  To   From   To     CAGR
0     0.1  1.07   1.42   -6.62%
0.1   0.2  1.42   1.75   12.99%
0.2   0.3  1.75   1.88   7.86%
0.3   0.4  1.88   2      10.19%
0.4   0.5  2      2.12   11.82%
0.5   0.6  2.12   2.34   19.59%
0.6   0.7  2.34   2.95   17.30%
0.7   0.8  2.95   3.508  10.73%
0.8   0.9  3.508  4.437  20.82%
0.9   1    4.437  6.63   18.04%
                         
1970                     
From  To   From   To     CAGR
0     0.1  1.07   1.52   -4.09%
0.1   0.2  1.52   1.82   11.34%
0.2   0.3  1.82   1.96   9.36%
0.3   0.4  1.96   2.11   11.68%
0.4   0.5  2.11   2.47   20.21%
0.5   0.6  2.47   2.98   9.16%
0.6   0.7  2.98   3.461  8.33%
0.7   0.8  3.461  4.05   13.54%
0.8   0.9  4.05   4.89   16.02%
0.9   1    4.89   6.63   20.02%
                         
1960                     
From  To   From   To     CAGR
0     0.1  1.07   1.624  -0.98%
0.1   0.2  1.624  1.87   10.33%
0.2   0.3  1.87   2.04   10.70%
0.3   0.4  2.04   2.3    16.23%
0.4   0.5  2.3    2.91   11.62%
0.5   0.6  2.91   3.09   3.85%
0.6   0.7  3.09   3.38   9.17%
0.7   0.8  3.38   3.76   14.06%
0.8   0.9  3.76   4.666  16.80%
0.9   1    4.666  6.63   18.60%
                         
1950                     
From  To   From   To     CAGR
0     0.1  1.07   1.69   1.41%
0.1   0.2  1.69   1.93   9.25%
0.2   0.3  1.93   2.14   11.96%
0.3   0.4  2.14   2.82   17.01%
0.4   0.5  2.82   3.05   4.04%
0.5   0.6  3.05   3.32   8.14%
0.6   0.7  3.32   3.72   13.31%
0.7   0.8  3.72   4.32   13.28%
0.8   0.9  4.32   5.3    17.77%
0.9   1    5.3    7.28   23.40%
                         
1940                     
From  To   From   To     CAGR
0     0.1  1.07   1.74   3.05%
0.1   0.2  1.74   1.99   8.97%
0.2   0.3  1.99   2.28   14.92%
0.3   0.4  2.28   2.97   10.07%
0.4   0.5  2.97   3.22   5.77%
0.5   0.6  3.22   3.62   11.07%
0.6   0.7  3.62   4.21   11.17%
0.7   0.8  4.21   4.83   17.10%
0.8   0.9  4.83   5.722  17.76%
0.9   1    5.722  8.71   21.81%
                         
1928                     
From  To   From   To     CAGR
0     0.1  1.07   1.79   3.89%
0.1   0.2  1.79   2.05   9.35%
0.2   0.3  2.05   2.79   16.60%
0.3   0.4  2.79   3.09   3.64%
0.4   0.5  3.09   3.4    6.96%
0.5   0.6  3.4    3.77   8.70%
0.6   0.7  3.77   4.29   10.41%
0.7   0.8  4.29   4.906  14.53%
0.8   0.9  4.906  5.81   13.72%
0.9   1    5.81   10.55  21.20%

Breaking up the data from 2000 to 2022 into twenty baskets also produces similar results:

2000                    
From  To    From  To    CAGR
0     0.05  1.07  1.24  -15.45%
0.05  0.1   1.24  1.37  -10.03%
0.1   0.15  1.37  1.48  -4.60%
0.15  0.2   1.48  1.66  6.97%
0.2   0.25  1.66  1.73  12.86%
0.25  0.3   1.73  1.78  10.62%
0.3   0.35  1.78  1.82  2.34%
0.35  0.4   1.82  1.85  8.97%
0.4   0.45  1.85  1.89  9.09%
0.45  0.5   1.89  1.92  8.88%
0.5   0.55  1.92  1.95  7.78%
0.55  0.6   1.95  1.99  10.68%
0.6   0.65  1.99  2.02  10.29%
0.65  0.7   2.02  2.05  8.28%
0.7   0.75  2.05  2.08  10.25%
0.75  0.8   2.08  2.12  12.79%
0.8   0.85  2.12  2.17  15.58%
0.85  0.9   2.17  2.23  15.97%
0.9   0.95  2.23  2.33  17.98%
0.95  1     2.33  3.58  23.71%

And I also did segments of history to see if the conclusions would be the same and they are:

1928  1950               
From  To    From  To     CAGR
0     0.1   2.85  3.44   -5.44%
0.1   0.2   3.44  3.70   -2.05%
0.2   0.3   3.70  4.12   1.82%
0.3   0.4   4.12  4.37   7.25%
0.4   0.5   4.37  4.72   6.35%
0.5   0.6   4.72  5.09   15.23%
0.6   0.7   5.09  5.78   -1.00%
0.7   0.8   5.78  6.41   10.32%
0.8   0.9   6.41  6.89   21.20%
0.9   1     6.89  10.55  28.02%
                         
1950  1990               
From  To    From  To     CAGR
0     0.1   2.63  2.99   -3.47%
0.1   0.2   2.99  3.13   3.45%
0.2   0.3   3.13  3.33   8.22%
0.3   0.4   3.33  3.56   12.32%
0.4   0.5   3.56  3.81   14.19%
0.5   0.6   3.81  4.19   11.94%
0.6   0.7   4.19  4.53   18.76%
0.7   0.8   4.53  5.12   15.89%
0.8   0.9   5.12  5.77   21.45%
0.9   1     5.77  7.28   24.25%
                         
1990  2024               
From  To    From  To     CAGR
0     0.1   1.07  1.37   -8.11%
0.1   0.2   1.37  1.64   7.25%
0.2   0.3   1.64  1.80   12.66%
0.3   0.4   1.80  1.89   7.72%
0.4   0.5   1.89  1.98   9.77%
0.5   0.6   1.98  2.06   11.14%
0.6   0.7   2.06  2.18   15.41%
0.7   0.8   2.18  2.41   20.82%
0.8   0.9   2.41  2.93   18.94%
0.9   1     2.93  4.09   14.81%

Well done. Thanks for sharing.

Interesting study Jim. I confess, I was highly skeptical.

Me too!
It didn't even dawn on me to check it till I noticed how remarkably flat dividend yield has been for a long time: other than the pandemic dip, basically flat since late 2002, the bottom of the tech bear.
If the yield is flat, and the price is going up and down, then it implies that the price is mean reverting to that flat number.
The implication is not on solid ground: it might not be a causal relationship, or might not be in that direction. But the effect is there in the last 20 years of data, anyway.

I didn't really look at what time frame it is the best predictor for. I just picked 18 months and tested it.
I did average of real total return series endpoints 1-2 years out, divided by its level today, ^(12/18)

It might be a stronger and/or more reliable predictor at 2-4 years, or other time frames.

Jim

Jim wrote: I didn't really look at what time frame it is the best predictor for. I just picked 18 months and tested it.

I've tested six, twelve, eighteen and twenty-four months. I'm seeing the same message no matter which I choose.

2000  6-Mths               
From  To       From  To    CAGR
0     0.1      1.07  1.37  -13.26%
0.1   0.2      1.37  1.66  1.88%
0.2   0.3      1.66  1.78  13.83%
0.3   0.4      1.78  1.85  8.28%
0.4   0.5      1.85  1.92  9.12%
0.5   0.6      1.92  1.99  8.85%
0.6   0.7      1.99  2.05  12.05%
0.7   0.8      2.05  2.12  11.95%
0.8   0.9      2.12  2.23  15.08%
0.9   1        2.23  3.58  24.23%
                           
2000  12-Mths              
From  To       From  To    CAGR
0     0.1      1.07  1.37  -14.59%
0.1   0.2      1.37  1.66  0.39%
0.2   0.3      1.66  1.78  13.89%
0.3   0.4      1.78  1.85  7.29%
0.4   0.5      1.85  1.92  9.36%
0.5   0.6      1.92  1.99  9.49%
0.6   0.7      1.99  2.05  9.61%
0.7   0.8      2.05  2.12  10.90%
0.8   0.9      2.12  2.23  13.35%
0.9   1        2.23  3.58  23.00%
                           
2000  24-Mths              
From  To       From  To    CAGR
0     0.1      1.07  1.37  -10.02%
0.1   0.2      1.37  1.66  3.82%
0.2   0.3      1.66  1.78  10.57%
0.3   0.4      1.78  1.85  2.35%
0.4   0.5      1.85  1.92  8.86%
0.5   0.6      1.92  1.99  10.35%
0.6   0.7      1.99  2.05  10.36%
0.7   0.8      2.05  2.12  10.62%
0.8   0.9      2.12  2.23  14.54%
0.9   1        2.23  3.58  20.15%

Jim wrote: The implication is not on solid ground: it might not be a causal relationship

I rather doubt it. If there is any cause/effect relationship going on here I would think it is the investor public seeking out yield wherever they can get it. When dividend yields are high they are happy to create demand for securities that offer that higher yield, and that demand in the context of similar supply drives prices higher.

To test this I took 10-Year Treasury rates as a proxy for what one might get for savings. Right now I'm getting 4.35% at Capital One 360. I'm also getting 5.10% from a 1-Year CD at the same bank. The 10-Year is presently 4.27. Of course, this doesn't always work out, but close enough for this test.

The theory is that when rates for safe investment options are good, money remains there safely tucked away in a bank or money market account. When the savings rate is very low investors seek yield elsewhere and are willing to take on more risk in securities.

So for the following table I break up the rate into deciles and baskets of twenty and show the S&P 500 return in each basket from 2000 to present.

2000                    
From  To    From  To    CAGR
0     0.1   0.70  1.74  4.88%
0.1   0.2   1.74  2.03  22.81%
0.2   0.3   2.03  2.34  12.94%
0.3   0.4   2.34  2.63  10.93%
0.4   0.5   2.63  3.04  15.66%
0.5   0.6   3.04  3.62  11.68%
0.6   0.7   3.62  4.11  8.94%
0.7   0.8   4.11  4.50  7.02%
0.8   0.9   4.50  5.00  -3.84%
0.9   1     5.00  6.79  -10.66%
2000                    
From  To    From  To    CAGR
0     0.05  0.70  1.52  8.87%
0.05  0.1   1.52  1.74  -1.51%
0.1   0.15  1.74  1.94  24.09%
0.15  0.2   1.94  2.03  21.08%
0.2   0.25  2.03  2.12  19.24%
0.25  0.3   2.12  2.34  10.36%
0.3   0.35  2.34  2.47  10.63%
0.35  0.4   2.47  2.63  11.25%
0.4   0.45  2.63  2.73  16.66%
0.45  0.5   2.73  3.04  14.76%
0.5   0.55  3.04  3.43  12.69%
0.55  0.6   3.43  3.62  10.52%
0.6   0.65  3.62  3.88  8.45%
0.65  0.7   3.88  4.11  9.60%
0.7   0.75  4.11  4.25  6.63%
0.75  0.8   4.25  4.50  7.42%
0.8   0.85  4.50  4.71  -1.43%
0.85  0.9   4.71  5.00  -6.28%
0.9   0.95  5.00  5.34  -7.53%
0.95  1     5.34  6.79  -13.83%

When you can make a good return in savings money is moved out of securities and into these safe investment options, which by definition, moves money out of securities causing demand to lessen and supply to increase dropping prices.

I'm more comfortable with this explanation, but I'm certainly open to others.

The theory is that when rates for safe investment options are good, money remains there safely tucked away in a bank or money market account. When the savings rate is very low investors seek yield elsewhere and are willing to take on more risk in securities.

It has some merit as an explanation, even if it isn't the most intelligent thinking for investors.

Perhaps strengthening this is the observation that the "flat 2% yield" effect I noticed started around 2002, with the noted exception of around the credit crunch when everything got a lot cheaper. The 2% yield notion certainly wasn't true in the last century. Essentially never below 2.7% at the lowest, before the tech bull started around '95.

By sheer coincidence, short bill interest rates have been pretty low since early this century, almost always under 2.4%.

Until recently. So maybe my lovely observation was posted right when it stopped being useful... : )

Jim

Jim wrote: It has some merit as an explanation, even if it isn't the most intelligent thinking for investors.

Well, I never claimed to be the sharpest tool in the box!

Jim wrote: It has some merit as an explanation, even if it isn't the most intelligent thinking for investors.
...
Well, I never claimed to be the sharpest tool in the box!

I think YOUR thinking was good, which was that many in the market might be thinking that way, so it has explanatory power for the observed numbers.
It's THEIR thinking that I think is a bit muddled : )
Reaching for yield rarely ends well, as people tend to lose focus on the underlying wealth generation (earnings) of the investment in question.

I would say you probably ARE the sharpest tool in the typical analysis box, but maybe even in this context it's not good to call someone a tool... : )

Jim

If we view long-run market returns as approximated by (GDPgrowth + inflation + diviYield) and long-run gov bond rates as approximated by (GDPgrowth + inflation) then dividend yield approximates the equity risk premium, the excess we can expect (hope) to get over safe investments. Not saying this is concrete, but maybe it's "roughly right"?

Perhaps investors realised that 1% yield (dot com boom) is not enough to compensate for market volatility, and have decided to stay closer to 2% since then?

SA

Where might one get a time series for dividend yield of an index?

I can construct a series from SPY using daily prices, but the dividend is paid out only quarterly. I suppose perfect would be to have the dividends of each S&P 500 component on the day they are paid out, but that's would be a lot of work. It seems Guru Focus has a daily data set (based on monthly div yield?) available for subscribers (which I am not).

Just wondering where people got their data sets, or how they constructed them.

Thanks,
John

John asked: Where might one get a time series for dividend yield of an index?

https://pinnacledata2.com/idx.html

* Dividend Yield of Dow Jones Industrials from 3/31/88
* S&P 500 Dividend Yield from 1/7/28

Both are weekly data and the pricing is very reasonable.

Dividends:
http://www.multpl.com/s-p-500-dividend-yield/table...

Prices from yahoo.

The dividend adjusted prices going back one rolling year are available at StockCharts to non-premium members and the public, one click from the chart. Scraping it at the non-premium level is not easy because it's not set up as a table.

Post New

Atheist Shrewds FAQ

---

Contact Shrewd'm
Contact the developer of these message boards.