Subject: Dividend investing
Emphasizing high dividend income when building a portfolio is, in a general sense, a bad idea. Usually.

Some people start with the throw-your-hands-up point of view that all stock market prices are pretty much random, so you might as well get a dividend--thereby falling for the myth that a dividend is an "extra" you get on top of stock returns. Not noticing that the stock price (and market cap) of a company drops each time a dividend is paid out. It isn't free money, it's just a different WAY of getting your money.

Then there's the observation that a high dividend is often a sign of a dead end firm - nobody is bidding the price up for a reason, and the dividend yield doesn't fall to a typical level. So a group of high dividend companies is at risk of being a group of dead end companies, cash cows at best, or headed for the graveyard. The average total return from a US stock with a dividend yield over about 8% is negative: the stock price drops on average at more than 8%/year.

But there are two good solid cases for a dividend-oriented portfolio. One obvious, one less so.

The obvious one is that your portfolio may be subject to tax laws that make a dividend very lightly taxed. This is often true, but of course it's often true that capital gains are also lightly taxed, so the difference may not be that big a deal. Many people would be better off ignoring dividends, buying the best companies they can identify, and (if needed) selling small amounts of stock from time to time for income. There isn't much more to be said about this because it depends so much on where you live, and where the company lives. For any subgroup the subject has been beaten to death elsewhere.

But, finally to the point of the post, the second and more hidden advantage of dividend investing is that dividends are not nearly as cyclical as stock prices. Not only does this mean that you don't get that much of a "pay cut" during a market downturn, but also that available dividend yields on new purchases tend to be correspondingly higher. This means that some turnover in a portfolio isn't necessarily a problem.

Why this post?

The outlook for US markets is not currently bright. I have no idea if the next year will be up or down, but the cyclically earnings yield is at the very low end of its historical range--you're not getting much in the way of earnings for your investing dollar--and that is by far the best predictor of poor returns over the next 5-10-15 years. Makes sense: in a way, all stock investing, whether growth or value, is about buying a stream of future earnings. If you pay twice as much for it, you get half the value in return.

Long story short, it might not be such a bad time for someone who has never been interested in a dividend-focused portfolio to consider whether it might make sense. If income from a portfolio is its prime function for you, the resilience of dividends relative to capital balances could be an advantage in the next few years.

So, to that end, here is a proposal for a way to have a portfolio with an unusually high dividend yield while (to the extent it works) avoiding the problem of high dividend payers that are bad investments on the capital side.

Quant investing to the rescue!

The portfolio I have been testing:
* Start with the set of 1700 stocks currently covered by Value Line. They are almost all American, and almost all bigger than microcaps, with a few exceptions to each of those.
* Find the 400 with the highest dividend yield.
* Of those, find the 80 with the highest reported ROE (Return on Shareholders' Equity), as calculated by Value Line from the most recent annual statements. This data field doesn't change very often, minimizing turnover.
That's the basic calculation for the nominee stocks.

The suggested portfolio management:
* Buy equal dollar amounts of the 50 stocks with the highest ROE. This is always a 50 stock portfolio, no more than 2% of your money put into any position, so in that sense it's only 29% of the risk of the S&P 500.
* Hold for three months.
* Recalculate the top 80 picks again.
* At the next trading date, sell any stocks no longer ranked among the top 80.
* Replace those with equal dollar amounts of the highest ranked ones (highest ROE) that you don't already own.
* Once a year, rebalance all positions to be equally weighted. (trim those that have gone above 2% of your portfolio, and top up those that are under 2%).

The end result is a portfolio that is a bit like a well built junk bond portfolio: a lot of the picks will be bad companies with poor prospects and a high dividend yield, but you're minimizing your losses by choosing from among those the firms with the highest ROE, which is the best single clue that a business actually has some good economic characteristics. You still get some losers, but they losses are more than made up for by the gains...and you get a very healthy dividend yield.

Historical results:
I built and started testing this screen a little over five years ago. In the backtest 2000-2019 it beat the market by quite a bit, 6.9%/year, allowing for 0.4% round trip trading costs. It has been basically market tracking in the five years since creation, which sounds bad, but the rate of return is within 0.1%/year of what it was in the backtest era, so in that sense it just keep chugging along.

Of more interest is the dividend yield.
I didn't calculate the yield for every date, but roughly monthly, with interpolation.
An average year since mid 2006 (19 years) had a prospective dividend yield averaging 4.30%. (prospective in that it counts neither the dividend increases nor dividend cuts you'd get in the first year of holding a stock--not an issue).
At times of low markets, the prospective yield is higher, and vice versa. The lowest prospective yield was in the 3.1%-3.2% range, hit a few times in 2008, 2009, and 2011. The highest forward yield for a date that I calculated it was 11.3% during the pandemic sell off.


I approximated the portfolio value and the dividend income separately since 2006.
The average rolling year saw the portfolio's value rise about 10.3%/year, and a dividend yield of about 4.3% on the portfolio value at the start of the rolling year. The cash dividend income rose at a rate of about 9.8%/year. The market has been strong (S&P total return 11.9%/yera), so I would not expect anything like that growth rate in future, but I would still expect a good solid dividend which also rises in real terms over time.

What sort of firms does it pick?
If you were starting this week, the prospective yield is 4.54%. The median ROE is 46.1%, lowest 33%. Median market cap $15.5bn: it mostly likes big companies. Some fairly well known firms in the top 50: Cisco, J&J, Coke, Amgen, Clorox, Pepsi, Nike, Kellogg, Albertsons, Whirlpool, Hasbro.

For those who like to improve quant screens: my advice in this case is “don’t”. I have tried innumerable variations of this trying to improve the overall returns or reduce the number of dud companies picked, but (a) nothing seems to work, (b) simplicity has its own resilient beauty, and (c) this specific simple version now has over five years of post-discovery data.

Jim