And the winner is ... dividends!

In Chapter 29 of Security Analysis, Graham and Dodd make the comment “Until recent years the dividend return was the overshadowing factor in common-stock investment”. (These words come from the 1940 edition of the book.)

To illustrate this they present two tables, one for American Sugar Refining Company (ASR) and one for Atchison Topeka and Santa Fe Railway Company (ATSR). Each table shows for a number of years, the price range of the stock, the earnings per share and the dividends per share.

ASR had quite volatile earnings, for example going from $18.92 per share in 1911 to $5.31 per share in 1912, but maintained a constant dividend of $7 per share over the period of the table (1907 – 1913). According to Graham and Dodd, the volatility of its share price was low, suggesting that it was the dividends driving the price rather than earnings.

The table for ATSR covers the years 1916 to 1925. The dividend paid was $6 per share each year except for the last when it rose to $7 per share. Earnings per share, while exhibiting some volatility, had a general upward trend. The price range again was relatively steady and level, but jumped in the final year when dividends were raised. So again, dividends seem to be driving the price.

I decided to undertake a similar exercise on a current ASX listed company. I chose National Australia Bank Limited (NAB) because I thought a long established bank would have steady dividends but volatile earnings and because, due to accidents of history, I own some (a very little some) NAB shares.

Instead of replicating the tables, I created two graphs: one plotting earnings per share against closing price on the day the full year results were announced and one plotting dividends per share for the year against closing price on the day the full year results were announced.

Here’s the graph of earnings against closing price for the years 2003 to 2007 inclusive. Closing prices were taken from Yahoo’s finance site, which only goes back to 2003. The R-square figure of 0.163 is a measure of how well the movement of earnings per share affects the closing price. A R-squared of 1 would mean that a change in the earnings per share would result in the exact same change in the closing price. A R-squared of 0 would mean that a change in the earnings per share would result in absolutely no change in the closing price. A R-squared of 0.163 is very low suggesting earnings per share has little influence on price.





What I see when I look at the graph is a squiggle – it folds back on itself. In this case I think the R-squared is meaningless and earnings do not directly affect the share price.

In contrast, when I look at the graph of closing price against dividends per share for the same period, things are a bit clearer. Each rise in dividends paid has lead to a rise in the share price. The R-squared of 0.664 suggests a reasonable correlation. So although there are not many data points here and no doubt a statistician would draw no conclusions from this, it would appear dividends directly affect share prices and earnings do not.




The impact of dividends could also be seen in the market reaction to Qantas’s decision to cut its interim dividend for the FY2010 half year despite the profit result being in line with company guidance. Its share price fell 8% in one day.

You may be wondering why I only took the graphs out to 2007. It’s because 2008 is when the global financial crisis hit and things went awry. When I extend the graphs out to 2009, it all goes horribly wrong.





Now both graphs are squiggles. So from this I conclude when it all goes wrong, it all goes wrong and hopefully when it happens again, I will have already switched my portfolio into cash.