What Pays Up Must Go Down
Ares Capital Corp. (ARCC) makes its money by providing loans to other companies and collecting interest on those loans. It’s a fascinating business and there’s a great argument to be made that it’s undervalued and worth buying for the long haul. But I’m not going to make that argument. Let’s say you don’t want to make “prudent investments” with “good prospects for long term returns” based on “sound company fundamentals”, let’s say you want money now. Well ARCC provides a dividend, and a pretty hefty one at that, hovering around $.40/share quarterly, which is over 11% annually as of 09/2020. 11% is substantial, but let’s assume that we investors are impatient and greedy people, and we want even more, even faster. Is there a way to take advantage of the dividend in some other way? Perhaps investors tend to sell just after they are assigned their dividends, creating an opportunity for us to buy in at a low price? This would increase the reward for us from all future dividends, or simply allow us to sell at a premium when the price returns to normal.
In order to test this strategy, it’s important to know the basics of dividends. A great source of information for this is investopedia, but in short, dividends are money that a company pays out to its investors. Dividends can be paid out for a variety of reasons, but they are most often given at regular intervals and considered part of the benefit of holding a particular stock. ARCC occasionally gives small special dividends, but we will be focusing on their much larger and more predictable quarterly dividends. After a dividend is announced, it is locked in on the ex-dividend date. You will only receive the subsequent dividend if you purchase the stock before that day. The money, earned by holding the stock on the ex-dividend date will be received on the dividend date, even if the stock is sold in the intervening time. If investors were going to leave a stock after taking a dividend, they would likely do so after the ex-dividend date. Let’s see if that holds true.
This plot shows the price of ARCC over time between mid-2015 and august 2020, with ex-dividend and dividend dates added as red and blue lines respectively. We are looking for changes that occur consistently on ex-dividend days.
It appears that there might be something to our hypothesis. The graph seems to reveal that price of ARCC often moves downward just after the ex-dividend date, but before the dividend date. Next we’ll attempt to capture a clearer view of these changes with a different type of graph.
This plot shows the change in price of ARCC from previous days on dividend days, ex-dividend days, and all days. By combining a violin plot and a box and whisker plot, we can see both the range and distribution of price changes on the relevant dates.
It appears that there is a consistent drop in price on ex-dividend days. In fact, it appears that on nearly all days, the price of ARCC either moves up or moves down by less than on the average ex-dividend day. Any day during which ARCC falls more than an average ex-dividend day would be an outlier.
So this is good news for our hypothesis right? Well, not exactly. While the price of ARCC does drop fairly consistently on ex-dividend days. It only does so by a very small amount. In fact, it’s often smaller than the dividend we miss out on by buying after the ex-dividend.
Here, using a complex form of math known only as “math”, we can see that whether comparing means or medians on ex-dividend days and all days, the difference in price is either smaller than the dividend, or greater by at most 1 cent. So, while you could buy the stock at a roughly $0.32 - $0.39 discount by buying the day after dividends are locked in, you would be missing out on… a $0.38 dividend.
This is not actually surprising. According to Vanguard, it’s common knowledge:
“On the ex-dividend date, the security’s price typically falls by the amount of the dividend. For example, if a stock was trading at $10 and declares a $.40-per-share dividend, the security’s price will likely fall by $.40, to $9.60, to account for the dividend.”
So was this a completely useless exercise? No. Good science requires confirmation yadda yadda yadda it’s important to ensure that research is reproducible blah blah, all data is good data, whatever; There is still a useful pattern here! While ARCC prices only fall by the dividend price on the day the dividend is locked in, on the day that the dividend is actually distributed, we find that they increase by a median of $.04! This is likely because many investors automatically reinvest their dividends into new shares, increasing demand for ARCC shares on days when dividends are parceled out. The difference in price is very small, amounting to only about .0025% of the average share price, but we can see that while there is no best day to buy ARCC, dividend days may, by a hair, be the best days to sell.
What really is the takeaway from this exercise? You should count every fraction of a fraction of a percentage point to ensure optimal returns on every stock? I don’t think so. To me, the real value here is confirming that, in some parts of the market, the efficient market hypothesis most certainly holds. Given crazy swings in markets, both recently, and less recently (looking at you Onion Futures Market Scandal of 1955), I find it hard to justify the belief that all available information is always baked into the market. Examining cases like this, however, shows that when something is as regular and predictable as dividends, the market has likely already accounted for it. Does this mean that there is no money to be made by timing ex-dividends? No, it almost certainly doesn’t. I would actually bet money that algorithmic traders are slowly amassing fortunes doing exactly that.
But it’s a waste of time for the average investor. Best not to fight financial gravity.