No. of Recommendations: 35

Some folks rely on dividend income.

Assuming you have a broad portfolio of dividend-paying stocks,

how much of a pay cut can be expected during this, the Great Recession?

How long will the drop last before going up to new heights?

Answer: nobody knows.

But dividends are certainly stickier than earnings.

Trailing-year broad US market earnings have dropped by 92.6% so far, to

$7.21, for a P/E of 138---an all time high. I estimate the sustainable

on-trend earnings level is somewhere around $52, though it could

be as low as $45 or as high as $65. A figure of around $52 implies

that the market is currently slightly *over*valued. But I digress.

US dividends on the other hand have historically never dropped anywhere

near as much. Using the Shiller database from 1871, this shows that

rolling-year dividends have never before dropped by more than 53.7% from their peak.

This is a fairly strong outlier from 1920 (not 1930).

A chart of the drawdown of dividends---current trailing year versus

peak trailing year so far---shows that a drop of 10% is common, having

happened around 14 times. Several times the drop bottomed around -12% to -15%,

for example 1905, 1910, 1954, 2002, and so far this time.

20% drops have happened nine times, but it's interesting to note

that these were not spikes---the drop lasted many years, probably

implying that the prior peak was unsustainable.

Six times the drop was in the 25-30% range.

Four times it was over 40%.

Bottom line? a drop of 50% overall isn't out of the question, and

is probably somewhere in the rough vicinity of a worst case based

on the last century and a half. Since trailing dividends have already

dropped about 12%, one might speculate that a pretty-much worst

case drop of half from the peak would involve a drop of 43% from here.

Ouch. But, if you need a yield of 4% and you buy a good diversified

portfolio yielding over 8% (not hard right now), you might just be OK.

But, this observation about drops from prior peaks lasting a long

time got me to thinking that dividends, like earnings, can have

unsustainable upwards spikes. Everybody knows we're in a low

dividend era, so this shouldn't be a problem, right?

Wrong. It turns out that we're in a low dividend *payout* era,

but real dividends themselves spiked way above trend in much the

same way as earnings did, though to a lesser degree. It's a heck

of a lot easier to get a good trendline fit through the very long

run real dividend line, so I built such a trend. The formula

comes out as 1.7425973689E-08 * e ^ (1.0412508300E-02 * year).

What does this tell us?

Well, by looking at the actual TTM dividends versus this trend,

you can look at the variance, and the standard deviation of the

variance, to get an idea of where we "should" be, and how far

above or below trend we are. Right now we are about .87 of

a standard deviation above the trend line, not too much---a mere

13% drop from here would bring us to the line of normality.

But, that's not very confidence inspiring, since an overshoot will follow.

By the way, the recent peak in dividends was at 1.89 standard

deviations above trend, which is why I comment that it seems to have been an unsustainable peak.

The graph of standardized deviation from trend shows (rather

axiomatically) that excursions down below -1 standard deviation

do happen from time to time--nine occurrences. Excursions of -2

standard deviations have happened twice, in 1920 and 1949, but

the worst was not much more than -2, at -2.14, and it rebounded quickly.

Interestingly, both times were double bottoms: dividends dropped to

-2.1, rose to -1.8, dropped to -1.9. It seems likely that the

current environment might well have a drop of a roughly comparable

magnitude, given how many things have gone wrong.

Dividends dropped very quickly 1918-1920, and rose fairly steadily

back to the trend line. In both the 1918-1920 drop and the 1939-1947 drop,

the only two times we had a -2 standard deviation drop, the time spent

below trend was between 9 and 10 years. They were similar in depth

and length, but the 1920 one was smoother, so I will use it as

a model to forecast what will happen starting from now.

The forecast based on deviation from trend would be as follows:

[all figures are "real" end-2998 dollars]

Current TTM real dividends $24.58, down from peak of $28.39.

What it would be now if we were on the long run trend line: $21.25

Forecast of 2009 total dividends (Dec 2009 TTM): $18.56, 85% of today's TTM

Forecast of 2010 total dividends (Dec 2010 TTM): $15.06, 69% of today's TTM

Forecast of 2011 total dividends (Dec 2011 TTM): $14.12, 64% of today's TTM

Forecast of 2012 total dividends (Dec 2012 TTM): $14.28, 65% of today's TTM

Forecast of 2013 total dividends (Dec 2013 TTM): $16.21, 74% of today's TTM

Forecast of 2014 total dividends (Dec 2014 TTM): $16.46, 75% of today's TTM

Forecast of 2015 total dividends (Dec 2015 TTM): $17.08, 78% of today's TTM

Forecast of 2016 total dividends (Dec 2016 TTM): $18.00, 82% of today's TTM

Forecast of 2017 total dividends (Dec 2017 TTM): $20.94, 96% of today's TTM

Forecast of 2018 total dividends (Dec 2018 TTM): $23.91, 109% of today's TTM

This is obviously not a line of reasoning that can be thought of as

valid or reliable, but I think it counts as plausible, or at least illustrative.

The underlying tenets are:

- Long run real dividends go up a certain average amount each year,

and are mean reverting to this trend line. Looking at the graph

since 1871, this certainly seems to be the case. This chart shows

the log real dividends and my forecast in yellow:

stonewellfunds.com/RealDividendForecast.jpg

- Two times in the past, dividends have dropped to 2 standard deviations

below trend, with the drop lasting about 9.5 years, so given the

circumstances the central expectation should be similar.

- It appears that though dividend yields have been low in recent

years, that's because prices were high, not because dividends were low.

It appears that, as with earnings, dividends were and remain

unsustainably high and further falls would be required (though not much)

just to get us down to trend.

- An overshoot below trend is extremely likely.

The resulting very tentative* conclusions are:

- Dividends might well drop in real terms by about another 40%

from their current trailing-year level in real terms.

This is maybe not quite as bad as some people were expecting, though

it means a peak to trough drop of somewhere around -53%.

The bad news is that it's very possible that the drop will be long-lived.

- The fall from peak to trough might be 2.5 years, with a bottom in mid 2011.

- The climb from trough to trend might be another 7.3 years, and the

trend should be very similar to the old high at that point,

so real dividends might hit both trend and new highs in 2018.

So, for financial planning purposes:

If you need $6 in dividend income, you'll need a portfolio yielding 10% today.

The standard Value Line edition has 78 stocks yielding over 8%, and 33 over 12%.

Hmmm, let's see---

Let's eliminate private equity firms, non-US funds, newspapers,

REIT's, and all stocks without at least a "B+" financial strength,

and all stocks without a current indicative dividend yield under 8.5%.

That leaves 22 stocks with an average yield of 10.1%.

(yes, there are still some income funds in there)

In case anyone is wondering, here they are with their prices, Financial Strength ratings, and current yields:

CAKE 13.35 B+ 14.1%

DTE 11.02 B+ 13.6%

USA 11.29 B+ 12.8%

WOR 15.12 B+ 12.4%

ABM 12.76 B+ 12.3%

CHRS 16.26 B++ 10.8%

JBL 24.42 B+ 10.7%

CLH 27.65 B+ 9.8%

CL 20 B+ 9.7%

IDA 12.86 B+ 9.6%

CHRW 30.19 B++ 9.3%

ADP 38.62 B++ 9.3%

NUAN 46.23 B+ 9.1%

FRX 24.14 B+ 8.9%

CREE 14.2 B++ 8.9%

TRAK 40.27 B+ 8.8%

FUN 32.43 B+ 8.8%

BIG 14.68 B+ 8.7%

TXRH 21.63 B++ 8.7%

ILMN 41.65 B+ 8.7%

APC 11.62 B+ 8.6%

BZH 39.96 B+ 8.5%

As dividends get cut on these, I would replace the cut ones with

whatever is closest to meeting these criteria at the time.

Voila! Income for life, with only a 45% pay cut to come.

I'd personally add in STD (Santander, the most solid European bank) and

WFC-PL (perpetual preferred shares from Wells Fargo yielding 10.8%)

Or, I suppose there's always the shotgun approach. Find all the VL stocks

that have yields over 4.75%. Assume that all yields over 15% are =15%.

Weight all the stocks by their yield, and you get a portfolio with

an aggregate yield of 10.0%---the largest stock holding is 3.16

times as large as the smallest stock holding and you hold 227 stocks,

so single-stock risk isn't exactly excessive at max 1.3% allocation per stock.

What will happen to all these stocks? I imagine half of them will

have big dividend cuts, and a lot of the other half will rise in stock price.

In both cases the current yield rate will fall. However, I imagine

the overall dividend yield as a fraction of the original portfolio

cost will not drop more than the forecast above (i.e., it will probably

stay above 6% of the initial portfolio value).

Jim

* = "belly laugh quality"