In The Wake Of Ex-Dividend (2024)

Investment thesis

There exists a price anomaly where the stock price would recover after the drop on the ex-dividend date. This recovery effect varies by holding period, and holding the stock longer (from 1 week to 4 weeks) produced a larger return with the same dispersion of returns. This anomaly is also exists after hedging with the SPY, showing the same pattern of increased returns for a longer holding period, with the same dispersion of returns.

Hedged returns can be considered as the stock outperforming the overall market, but there is no relationship between the stock's beta and the returns. This suggests that this price anomaly is a source of alpha.

Are there so many price anomalies around dividends?

I have previously written about 2 price anomalies around dividends: the price tends to rise between declaration date and ex-dividend date, and ex-dividend date price drop tends to be less than the dividend amount. There are anecdotes of yet another price anomaly a little further down the timeline, which is after the ex-dividend date price drop (or rise, if that happened). This is a curious little anomaly which I have not observed in my own trading, but have only seen mentioned about in various sites, and it is only mentioned in passing without much detail provided.

While my article on dividend stripping (or dividend capture, if you prefer the term) tried to isolate the price anomaly to only the overnight price drop, I also noted that there were other variations of dividend stripping advocated by other websites. A number of these included holding the stock for some time after the ex-dividend date, until the stock price recovers some (or all) of the ex-dividend date drop. If the price does recover after the ex-dividend date, this recovery is an anomaly we can make positive alpha from.

Investigating and separating the anomaly

In my humble opinion, the price movement after the stock goes ex-dividend is a separate price anomaly from the overnight drop. I could not find any studies done on this price anomaly (other than a single article on a single stock), and academic research is scant in this area with most studies focusing only on the ex-date. The various internet sources also do not give any good gauge on the recovery time period, with Wikipedia being the most specific when it mentions "a matter of weeks".

Unsatisfied with the lack of data and evidence about this anomaly, I set out to do my own analysis. While individual stocks may or may not all move as expected, I am trying to find out whether this anomaly exists at the aggregate level across many stocks and many ex-dividend dates. Being consistent with my previous investigations on dividend related price anomalies, I will use the 30 stocks in the Dow Jones Industrial Average, and investigate from an arbitrarily selected start date of 01 Jan 2001 (01/01/01) to present (12 Sep 2019 as of the time of writing).

The ex-dividend date price drop is the drop from close of the day before the ex-dividend date, to the open of the ex-dividend date. Thus the recovery can only take place after that, and I will start the investigation from the time the stock opens on the ex-dividend date. I also have no idea how long the price recovery period should be, so I will try examining the stock returns over the periods of approximately 1, 2, 3, and 4 weeks (5, 10, 15, and 20 trading days respectively). Dividends with the ex-dividend date in the past 4 weeks (from the time of writing 12 Sep 2019) will be excluded from the analysis, because obviously you cannot calculate a 4-week holding-period return when 4 weeks have not even passed.

The below timeline graphically represents what we are investigating:

(Created by author)

Examples using single stocks

The third-recent dividend in my data set is the BA dividend, with the ex-date of 08 Aug 2019. (The two most recent dividends are XOM and AAPL, with ex-date of 12 Aug 2019 and 08 Aug 2019 respectively, but I will use them later in the hedging example.) BA's opening price on 08 Aug 2019 was 331.30. The below chart shows the price movement since the ex-date:

(Created by author using chart from Investing.com, and data from Yahoo! Finance)

Another dividend on the same day is the WMT dividend (with the same ex-date of 08 Aug 2019):

(Created by author using chart from Investing.com, and data from Yahoo! Finance)

Both WMT and BA showed the same pattern, a negative 1-week return, and positive 2-week, 3-week, and 4-week return, with the returns increasing as the holding period increases.

Aggregate result of stocks in the Dow Jones Industrial Average

Repeating the above analysis on all the dividends of all the stocks, below are the aggregate results for the 30 stocks in the Dow Jones Industrial Average, with the results being the sum of individual trade returns, from 01 Jan 2001 to present:

(Created by author using data from Yahoo! Finance)

I have to emphasize once again that the above results DO NOT represent annualized return, CAGR, or even total return. The magnitude of the numbers is irrelevant, and only whether they are positive/negative indicates the existence and persistence of a price anomaly. The numbers DO NOT represent the back-test of any strategy.

The aggregate results generally show the same pattern that we observed from the BA and WMT examples above. Returns seem to increase as we hold the stock for longer periods of time. The dispersion of returns measured by the standard deviation also remained at around the same level as holding period increases (from 2 weeks to 4 weeks). While our BA and WMT examples showed a negative 1-week return, at the aggregate level the 1-week return is positive.

It could be argued that in the earlier examples, the stock prices recovered because the overall market went up. After all, both charts showed similar patterns and similar uptrends toward the end. Did the overall market go up? And can the market effect be stripped out? That's when we turn to hedging.

Do price recover because of, or in spite of, the overall market?

We will first look at 2 examples, and then the aggregate results. We first look at XOM, with ex-dividend date of 12 Aug 2019, comparing it with the movements of the SPY which approximates the overall market (XOM on the left, SPY on the right):

(Created by author using chart from Investing.com, and data from Yahoo! Finance)

Next is AAPL dividend with ex-dividend date of 09 Aug 2019 (similarly AAPL on the left, SPY on the right):

(Created by author using chart from Investing.com, and data from Yahoo! Finance)

The case of XOM showed negative returns in the first 3 weeks, and then a recovery in the fourth week, the recovery seemed driven by the rise in the overall market (represented by the SPY). The XOM stock price recovery after the ex-dividend date appeared to be because of the overall market.

AAPL on the other hand, showed that the price has already recovered in the first week, despite the overall market going down, with the fourth week return increasing even further when the overall market rose. The AAPL stock price recovery after the ex-dividend date appeared to be in spite of the overall market.

The following table shows data at the aggregate level, with the results being a sum of individual hedged trade returns, from 01 Jan 2001 to present:

(Created by author using data from Yahoo! Finance)

As expected, hedging reduced the absolute amount of returns. However, the same pattern is still seen, which is returns seem to increase as we hold the stock for longer periods of time, despite hedging out the effect of the overall market. The dispersion of returns also remained the same despite the longer holding period.

Alpha or Beta

Remember that hedged return can also be understood as the stock outperforming the overall market. And given that the overall market tends to drift up, could the positive hedged returns be explained by the stocks having high beta? If this were the case, we would expect stocks with beta greater than 1 to have higher hedged return (or outperform the market more), and stocks with beta less than 1 to have lower hedged return, or even negative hedged return. The following table presents the hedged returns compared with the stocks' beta extracted from Bloomberg (field Beta: M-3):

(Created by author using data from Yahoo! Finance and Bloomberg)

It can be seen that beta being greater or less than 1 has no effect on the hedged returns. The correlation of hedged returns with beta is also negligible in the 3-week and 4-week return cases. This suggests that after the ex-dividend date, the stock price recovery is not beta, but an alpha. This price anomaly is a source of alpha.

Number of days for price to recover

Lastly, I have seen some websites recommend stocks for dividend stripping (or dividend capture) based on the average number of days it takes for the stock price to recover to the level before the ex-dividend date. I thought I might do a similar analysis here. In analyzing speed of recovery, I found a few points that might be important to note. Firstly intra-day prices are considered, as long as the high of the day is equal or greater than the close price before the ex-dividend date, it is considered a recovery because the investor with a limit sell order will get his order filled. It does not have to close above the price before ex-date.

Secondly, there are some cases where the stock price took extremely long to recover. For the INTC dividend with ex-date 04 Feb 2004, it took more than 10 years for price to recover to the level before the dividend (recovered on 11 Jul 2014). As such, the average days figure will be skewed by these cases.

Thirdly, the number of days it takes to recover is heavily influenced by the general market trend. Prices recovered quickly when the market is heading up, and slowly if the market is heading down. In the extreme cases like the INTC dividend, a down-and-sideways move led to an extremely long recovery time, skewing the analysis.

Nevertheless, here are the figures for those who are interested:

(Created by author using data from Yahoo! Finance)

Despite the severe limitations of using the average number of days as an analysis method, it can be seen that stocks with no alpha (negative hedged returns) generally have longer average recovery times. Or conversely, a short recovery time can be a source of alpha. However, remember that these are long-term averages across many dividends from as early as 2001, and that the recovery time can vary significantly even among the same stock's dividend.

Conclusions

The anomaly of a price recovery after the ex-dividend date exists at the aggregate level. The recovery increases the longer you hold the stock after the ex-dividend date. Even after attempting to remove the effect of the overall market by hedging with short SPY, this effect still shows as aggregate positive hedged returns for longer holding periods. This price recovery anomaly is also not linked to the stock's beta, and can be said to be a source of alpha.

The number of days for the stock price to recover to the level before the ex-dividend date, has some limitations as an analysis method, and I personally would not put too much weight on that set of results, even though it agrees with the rest of the data.

It is also possible that the price recovery takes longer than 4 weeks to complete, and that holding the stock for even longer might produce higher returns. However, my opinion is that it would be too similar to traditional buy-and-hold investing. If you decide to hold a stock for 2 months (or 8 weeks) after the ex-dividend date, for a quarterly-dividend stock, you are only 1 month away from the next dividend, where you are going to have to buy and hold it for another 2 months. The net effect is that you are holding the stock two-thirds of the time compared to a buy-and-hold investor.

Now imagine further that you want to combine this with the other 2 dividend-related price anomalies (run-up effect and dividend stripping/capture), and that the quarterly-dividend company declares dividends a month before going ex-dividend. You will be buying the stock a month before the ex-date, holding it across the ex-date, and holding it for 2 months after the ex-date. Around the time when you intend to sell, the company would declare another dividend, and you would hold it for the next cycle. Voilà, you have just turned into a buy-and-hold investor! Sure, quality companies are worth buying-and-holding, but then you wouldn't need me around to investigate all these price anomalies.

Proprietary trader for 8 years, traded in a wide variety of markets, varied trading style, extensive research experience, quantitatively inclined, favors systematic trading strategies.