If it didn't drop by the amount of the dividend, then people would be making an awful lot of money purchasing stock a day before the ex-date, being the owner of record, collecting the dividend, and then just selling the stock right away.
Well I just looked through tons of dividend distributions and stock price doesn't drop by the amount of the dividend in like 95% of cases. I don't dispute the theory about why you are right, it absolutely makes sense, but there's a mountain of evidence that it doesn't happen in reality.
For most companies the dividend distribution is less than average daily variance of stock prices. Which is why your strategy won't necessarily yield great results. A 3% yield distributed four times a year is only 0.75% of the stock value.
The SEC also has special rules for extremely large dividends so that the ex-date is after the distribution date.
I don't know how hard did you look, but taking for example the two stocks that you mentioned (MSFT and PG) and the ex-dividend dates since 2007 it's easy to see that there is an effect. The mean return for those dates is -0.5% for MSFT (vs. +0.05% for the market) and -1.0% for PG (vs. +0.1% for the market). Doing a regression of stock returns on market returns you will find that the result is statistically significant (p<0.001). R code follows:
Do you happen to have any good resources for learning R? I've always put it on the back burner but it looks like it has some amazing libraries for extracting data and now I'm keen to look into it more.
It's entirely possible that either the stock or market may be volatile, but that's in addition not, instead of, the drop in value resulting from a dividend payout. So, if we have a $100 stock with a 3% yield paid 4x a year, then the stock will drop $0.75 on the ex-date. Let's say, for the sake of argument, that you have some combination of market/stock volatility that moves the stock $4 in either direction. So, in the case of the stock going up $4, on the ex-date, it would go up $3.25. In the case of the stock going down $4, on the ex-date it would go down $4.75.
Yes - it's true that if you are just looking at a stock chart, it might be hard to eyeball the date of the dividend if there is a lot of volatility, but, presuming that volatility is evenly distributed over various stocks dividend dates, there is no way to avoid having the stock drop down by the value of the dividend. I guarantee you there is a pretty large number of investment bankers who have all sorts of models that arbitrage that kind of variance, pretty much eliminating any gap, should it exist for more than a few seconds after opening.
I don't understand how the ex-date could occur after the distribution date. The whole idea behind the ex-date, is that is the date on which the dividend belongs to the seller, not the buyer. I.E. If I buy the stock on (or after) the ex-date, I do not get the dividend, it belongs to the seller. Depending on the exchange, it's usually two business days prior to the date of record set by the board of directors. (At least on the SGX, the exchange I'm familiar with). So how could you place the ex-date after the distribution date? What would the date of record be in that case?
