Monday, October 29, 2012

Defending ‘Homemade Dividends’

Dividend investors and indexers often disagree strongly on the relative merits of their investing strategies. Recently, the Dividend Growth Investor argued that homemade dividends produced by selling some stock are not as good as real dividends. However, we can easily show that the core of the disagreement comes down to whether or not dividend stocks have an expectation of higher total returns.

For the purposes of this discussion, let’s compare an indexed portfolio of stocks that pay a 2% dividend to a dividend stock portfolio that pays an average of 4% dividends, both in tax-advantaged accounts. For the investor who wishes to live on 4% of his portfolio each year, his choices are to go with the indexed portfolio and sell 2%1 of his shares each year, or go with the dividend portfolio and live off the 4% dividend.

Dividend Growth Investor argues that “when someone sells a portion of their portfolio, they end up with less [sic] shares.” However, if the two portfolios get the same total return, then the shares in the dividend portfolio will be worth 2% less than the index portfolio shares are worth just prior to the sale of 2% of the index shares. After this sale, the two portfolios will have the same portfolio value.

Dividend Growth Investor goes on to argue that “Sometimes share prices fall or stay flat for extended periods of time, which could spell trouble for these [index] investors.” The implication here is that the index portfolio is shrinking, but the dividend stock portfolio is not. However, this can only be true if the dividend stocks are getting a higher total return. Otherwise, if index stocks are flat, dividend stocks are suffering a 2% capital loss each year.

I could go on, but all of the arguments come down to the same thing: will carefully-selected dividend stocks get higher total returns than the index or not? If the dividend stocks outperform, then the dividend investors are right; otherwise the indexers are right. Unfortunately for the dividend investors, the apparent evidence for long-term dividend stock outperformance has significant survivorship bias. If the indexers are right, then they have the advantage of a slight boost to returns that comes from better diversification.

1 The actual percentages are slightly off these approximate figures due to compounding effects, but I’ll use the round numbers to avoid muddying the waters.


  1. I think this analysis misses an important factor -- the time series of returns. If you're leaving an account alone, time series doesn't matter, but if you're pulling 4% a year out of the account, time series can become very important.

    The market price of a stock can go down and stay down for a long time. If you happen to need to pull a fixed amount of money out of the portfolio while stock prices are depressed, you could end up selling a significant percentage of your holdings over the course of a few years. Even if prices then rise, if you've sold 25% of your original holdings over a three year period, you've still experienced a disaster.

    Dividends of course can be cut as well. There are no certainties. But if you look at Canadian banks, just to use an example, their dividend history is a lot more stable than their share price history.

    I think your argument holds water if the investor has the flexibility not to sell stock for a few years. Maybe you have other assets to draw on, or a large enough cash buffer. But if you need to draw a fixed dollar amount from a portfolio like clockwork, I don't think it does, or at least I don't think it does with a high degree of certainty.

    Focusing purely on expected long term returns, while ignoring time series of returns as a factor is a frequent weakness in these theoretical strategy A vs. strategy B debates. Selling some portion of an indexed portfolio may indeed give you higher expected returns over a given time period, but I think it also gives you a higher variability of returns, and that variability matters a lot if your rent depends on it the returns.

  2. @Matt: I agree with your comments. However, as far as I can tell, they apply equally to both index-based portfolios and dividend stock portfolios. If one portfolio has an advantage, it can only be because there is something different about the returns of the portfolios; they must have different distributions. As far as I can tell, all of the arguments dividend investors make all boil down to believing that dividend stocks give better total returns than the index. For example, your observations about Canadian banks boils down to an expectation that Canadian banks will give total returns with lower volatility or higher expected total return than the general market.

  3. Great analysis Michael but I expect nothing less from a math guy.

    In the end, it's total return that matters and better still, it's real returns that matters.

    My strategy employes both dividend stocks and indexing, so I get to play on both sides of this argument like a two headed monster. It is Halloween soon after all - right? :)

    I recall most longitudinal market studies have demonstrated dividends represent a significant source of overall returns, which means for an investor like me, steady income. Do I take a lower yield and more appreciation potential (no dividends) or take higher yield and less appreciation potential (dividend stocks)?

    The beauty is, investors can do both.


  4. How about investing in ETFs that track dividend paying stocks? Eg CDZ, XDV and the BMO one? Any drawbacks here? You are indexing while still generating dividends.

  5. @Mark: The overall effect of mixing the strategies is to give your portfolio a value tilt. As long as you own enough dividend stocks, you shouldn't suffer too much from a lack of diversification.

    @Be*en: Using dividend ETFs gives decent diversification, but omits a large percentage of stocks. So, I wouldn't exactly call it indexing, but it's not too far off.

  6. Thanks Mike.

    If one is nearing retirement, would you recommend a shift from a "broad" index ETF to a "dividend" ETF? The latter I think would be less volatile and there's the dividends which get "dripped" until required as income.

  7. @Be*en: If I did give investing recommendations, I don't think you should follow them; it's better to make up your own mind. I'm unlikely to follow the strategy you describe. I'm likely to choose some mix of stocks, bonds, GICs, and cash, and just sell enough to live on each year. But if you're right about dividend stocks having the lower volatility going forward, then your strategy may work well. I don't plan to count on dividend stocks having better return distributions than the general stock market.

  8. Thanks Mike for your prompt reply. It is sometimes necessary to hear differing views to find one's bearings!

  9. I see the other huge problem with the Dividend Growth Investor . That is "carefully-selected dividend stocks", meaning that you actively managing your portfolio all the time.
    Particularly when you are retired....If it would be so easy to select the winners, everybody did it.

    More over it is far more riskier strategy, while promising very little potential reward.

  10. @Financial Independence: You make a very good point. The time that must go in to managing an active stock-picking strategy has to be accounted for unless you really would rather do this than anything else.