I have been using Beta in my dividend stock screening process, as a way of minimizing risk from volatility, for quite a while but recently I have had doubts whether it provides value. Part of the problem is that I found Beta to be confusing: not the definition, but how to use it properly. I was not clear about how well Beta is correlated with performance measures such as Dividend Yield and Total Return, especially in Bull vs Bear markets and for different market sectors. This paper is an amateur attempt to better understand how Beta behaves in different situations and where is the best place to use Beta. I am not a trained financial analyst, so there are parts that may be off base, but I do understand statistics.
What is Beta?
First, a reminder, what is Beta? A stock's Beta is its relative volatility compared to a market index. By definition, the market index has a Beta volatility of 1. A stock with a Beta <1 has a lower volatility than the market index and a stock with a Beta >1 has a higher volatility than the market index. Stocks can have a negative Beta if their price moves in the opposite direction of the market index. The calculation of a stock's Beta looks at its price movement relative to the market index through regression analysis over a fixed time period. A standard time period and comparative indexes are not necessarily defined (different stock market reporting websites use different Beta calculations), so some investors prefer to calculate their own Betas.
There are many stock investors, like me, who are looking for high dividend income while minimizing the risk of capital loss. Some of us look at Beta as a risk indicator and a stock with a Beta of less than 1 is often preferred. It is a risk avoidance strategy that is more concerned about the market falling than rising. It appeals to investors who care more about protecting capital than capturing stock market price gains.
How I use Beta as a predictor of dividend stock value?
I use Beta to predict "dividend stock value". Dividend stock value (DSV) is my own term. Others may have the same name for something else, but I define it as a metric to evaluate and rank dividend stocks as investments for my two broad objectives of 1. maximize Dividend Yield and 2. minimize capital loss risk. I use the following prediction equation:
The objective is a high (DSV). I usually have a threshold of (DSV) > 2 in my stock screening. The equation favours stocks with a high (DY) and low Beta, (PB) and (DE). Since Beta, (PB) and (DE) are all measures of risk, then the lower those numbers the better.
Why use Beta and Concerns
One anomaly with using Beta as a risk indicator is that stocks with higher dividend yields tend to have lower Betas. You might expect that stocks with high dividend yields should have higher risk, but that does not appear to be the case. This is illustrated by Dividend Anchor Score which was proposed by Frank Weiler in his book (Insync Income: the Must Read Guide to Investing in Canada. 2011). Dividend Anchor Score equals: Dividend Yield divided by Beta. Weiler hypothesized that "dividends act as anchors for share prices, thereby reducing volatility, and that dividends that are less likely to be reduced or interrupted make better anchors." If that is true, then it follows that good high dividend yield stocks will have lower a lower Beta, and there can be an inverse correlation between Dividend Yield and Beta.
I am having second thoughts about using Beta in my DSV equation. When I started thinking about what Beta really is, I started wondering if it is a legitimate predictor variable. The concern is this: Beta is a dependent variable. It is not intrinsically related to the value of a stock in the same way that company performance data independent variables are; e.g., earnings, profit, debt and equity. These independent variables can legitimately be used in prediction equations. Rather a stock's Beta is a function of everything about the company and the market, i.e., it is dependent on everything else. So statistically, it is questionable practice to use a dependent variable like Beta as an independent variable in downstream prediction equations. However, that doesn't stop us from looking at correlations between Beta and various stock metrics. Correlation does not imply "cause and effect" in the same way that prediction equations do.
Correlations between Beta and Dividend Yield and Total Return
Given the above concerns, I decided to look at the correlations between Beta and various dividend stock performance metrics, especially total 1 year return and dividend yield.
Here is how I did the analysis.
1. I used the S&P 100 as my database. The advantage is that it covers all the major US blue chip stocks and most dividend investors probably own several stocks in the S&P 100.
2. There are 8 stocks in the S&P 100 that do not issue dividends and I removed them from the analysis.
3. I looked at two time periods:
- June 16, 2014 to June 15, 2015 which was a "bull" interval where the total 1-year return was 10.4%
- August 25, 2014 to August 24, 2015 which was a "bear" interval where the total 1-year return was -2.8 %. Do you remember the market close on August 24, 2015? It wasn't pretty!
4. I used the Globeinvestor website (www.globeinvestor.com) for stock numbers: price, dividend, Beta, total return, etc. The period of total return (stock price change plus dividends) is one year. The Beta period is not defined, but that is probably not important as long as all the stocks in the index have a Beta measurement over the same period.
First of all, how good is Beta as a risk indicator? There should be a positive correlation between 1-year total return and Beta in a bull market and a negative correlation between 1-year return and Beta in a bear market. Here are the charts.
Here is total 1-year return versus Beta for June 2015 (the bull interval).
There is no correlation between total 1-year return and Beta. This does not support the assumption of: greater risk yields greater reward. Average 1-year return was 10.4%.
Here is the same figure for the August 24, 2015 (the bear interval).
The correlation is better, but it is still not great. Generally, the stocks with higher Beta lost more value in a bear market, which is what "risk-reward" theory would predict. Average 1-year return was -2.8%.
Turning now to dividend yield vs. Beta, you can make the point that dividend yield is more predictable than total 1-year return, so hopefully the relationship between Beta and Dividend Yield will be stronger than the relationship between 1-year return and Beta. Here is Dividend Yield versus Beta for the June "bull" and August "bear" periods.
Here, the pattern is similar for both the bull and bear periods. The negative correlation between dividend yield and Beta is consistent with the theory of Dividend Anchor Score.
One question would be, does the relationship between Dividend Yield and Beta hold for the different market sectors? Here is the total 1-year return and dividend yield analysis for the different sectors of the S&P 100 for both the bull and bear intervals.
Table 1. Correlations (NYSE:R) of Beta with total 1-year return and dividend yield for different sectors of the S&P 100
|Total 1-year return vs. Beta||(DY) vs. Beta|
|Sector||N||Jun return||Jun R||Aug return||Aug R||Jun R||Aug R|
No correlations were calculated for the Materials, Telecommunications and Utilities sectors due to low N
Things to note from Table 1:
1. In June only one sector (Energy) showed a decline in 1-year total return. In August, 4 sectors showed a decline.
2. There is a much stronger risk-reward relationship between Beta and total return in the bear vs. bull market.
3. The Anchor Score relationship (negative correlation between Dividend Yield and Beta) holds for all sectors except Industrials
4. There is no difference in the Anchor Score relationship in bull vs. bear markets.
5. There are better correlations for some sectors than others and versus the whole market. Beta correlations are strongest for Consumer Discretionary, Financial and Industrial.
Here are some of the charts illustrating the correlation between Dividend Yield and Beta for some individual market sectors.
Note that for the financial sector, DY is negatively correlated with Beta but for the Industrial sector, (DY) is positively correlated with Beta. Overall the correlations are quite weak and there are only 14 stocks in each sector, which is minimally adequate for correlation analysis.
Here is an example of a good sector correlation in the bear market interval.
I don't want to overstate the importance of the above Energy chart. All this means is that energy stocks were hurt badly the past year and the greater the Beta, the more they were hurt. This is obvious, and conforms to the risk-reward rule.
Dividend growth is considered by some investors to be important since it gives an indication of the safety of a stock's dividend. Here is dividend growth rate versus Beta for the June bull market interval.
The above chart shows a positive correlation between dividend growth rate and Beta, which is the opposite of the dividend yield to Beta relationship. Higher dividend growth is associated with more volatility. This may simply mean that, in a bull market, that stock price responds to dividend increases. The correlation is weak, though.
So it is appropriate to look at the correlation between dividend growth and dividend yield.
It is clear that low dividend yielding stocks show a greater percentage increase in dividend growth than high dividend yielding stocks. This is not surprising and explains why dividend growth is perhaps not the best metric of company dividend stability due to its bias against high dividend yielding stocks.
Finally, let's look at Total 1-year return versus dividend yield. One year return is not correlated with Beta but Dividend Yield is negatively correlated with Beta. It might be expected that total 1-year return will be negatively correlated with Dividend Yield as we expect high dividend yielding stocks to have a lower Beta and therefore a lower return in a rising market. As per the graph below, that is the case. There is a negative correlation between Total 1-year return and Dividend Yield.
1. Overall, I conclude that Beta is not that important in picking dividend stocks. Accordingly, I am dropping it as a measure of Dividend Stock Value in my stock screening prediction equation.
2. Beta does have some subjective value in looking for good dividend yielding stocks and generally obeys the Anchor Score theory in which there is an inverse relationship between Dividend Yield and Beta risk. Look for high Dividend Yield and low Beta but pay attention to the sector. Beta has highest value as an indicator in the Consumer Discretionary, Financial and Health Care sectors. I probably will be reluctant to pick stocks in sectors with a positive Dividend Yield - Beta relationship such as Industrials.
I probably could have done a more comprehensive analysis, e.g. use S&P 500 instead of S&P 100 as the database, but this is good enough for me. I have done enough analysis to make a conclusion but not enough to write a thesis. Nobody pays me to analyze.
That is all I have to say about Beta. Next I am going to look at my (DSV) prediction equation in terms of strengthening it to account for earnings, which I think will be important in minimizing the risk of capital loss. I may also revise the equation to reduce the emphasis on debt. Both (DE) and (PB) are debt related measures. I probably only need one term in order to not overemphasize debt. Debt probably shouldn't be given extra importance in these low interest rate times that we find ourselves in. Another problem with the (DSV) equation is that it does not eliminate enough stocks from consideration -- see footnote.
Footnote: Impact of dropping Beta from (DSV) formula
Using the August 24th database, the old (DSV) formula (with Beta) picked 30 out of 92 stocks in the S&P 100 which met the minimum (DSV) of 2. Excluding Beta from the formula dropped 7 of those stocks: Eli Lilly (NYSE:LLY), General Dynamics (NYSE:GD), Merk (NYSE:MRK), Microsoft (NASDAQ:MSFT), Nike (NYSE:NKE), Raytheon (NYSE:RTN) and Walmart (NYSE:WMT). Excluding Beta picked up one new stock: Abbott (NYSE:ABT).
Using the old (DSV) screen the top 3 stock picks in order were: Chevron (NYSE:CVX), ConocoPhillips (NYSE:COP) and Southern (NYSE:SO). Using the new (DSV) screen the top 3 stock picks in order are: Chevron, Exxon Mobil (NYSE:XOM) and ConocoPhillips. Note that they are mostly oil companies. All have good valuations and balance sheets but it goes to show that my (DSV) formula doesn't tell the whole story. I don't know too many people who are rushing to buy oil companies now.
Disclosure: I am/we are long "CVX", "WMT".
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.