Bad Idea: Playing The Mortgage REIT Dividend

|
Includes: AGNC, ARR, CIM, CYS, DX, IVR, MTGE, NLY, TWO
by: James Shell

We have been doing a lot of work on a little population of Mortgage REITs lately, and the following question was asked by one of my followers: What if you bought one of these funds shortly before the issuance of the dividend, and sold it shortly afterward?

This is an important question for several reasons: First of all, it suggests an easy way to make money, which is something of interest to many of us. Secondly, it is a fundamental principle of finance, that has applied from the time of the development of corporations until mid- 2008, that holders of capital should be rewarded for deferred consumption and not for trying to make a fast buck. Finally, since the mREITs have unusually high dividend payouts in this era, they might make a good laboratory for such a thought experiment since they are relatively less influenced by the noise that happens in the marketplace that might upset the calculation.

To recap, here is our universe of mREITs:

AMERICAN CAPITAL AGENCY CORP AGNC
ARMOUR RESIDENTIAL REIT INC ARR
CHIMERA INVESTMENT CORP CIM
CYS INVESTMENTS INC CYS
DYNEX CAPITAL INC DX
INVESCO MORTGAGE CAPITAL INC IVR
AMERICAL CAPITAL MORTGAGE MTGE
ANNALY CAPITAL MANAGEMENT INC NLY
TWO HARBORS INVESTMENT CORP TWO
Click to enlarge

The recent chart of many of these funds has been flat-to-negative:

Click to enlarge
(Click to enlarge)

The natural order of things is for the price of a security to drop immediately after the Ex-Dividend date, which is usually 3 days before the "Shareholder of Record" date from the dividend announcement, since this is the last day that a prospective shareholder can buy the security and expect to receive a dividend. Here is an example:

It is not It is not an absolute certainty that this happen, though, as the following example illustrates:

So conceptually, the idea of playing the ex-dividend date might still work if the dividend is big enough to offset the decrease in the stock price or, if you do it every time and play the odds that it will succeed more times than it will fail.

To test this I constructed a big spreadsheet with all of the stock prices and dividend dates. I will be happy to forward this to any of my followers who are inclined to review the conclusions or search for another way to play the game.

Buy the day before the Ex-Dividend date, Sell the day after, Collect the Dividend
Symbol Success Fail Even Proceeds Inv Per (Days) Annualized ROR
AGNC 4 4 0 ($7.20) 646 -15.2%
ARR 4 4 0 $0.10 290 1.7%
CIM 4 4 0 ($0.30) 651 -4.1%
CYS 4 4 0 ($12.70) 638 -55.7%
DX 4 3 1 $0.20 653 1.2%
IVR 4 4 0 $0.00 644 0.0%
MTGE 0 1 0 ($5.00) 100 -103.5%
NLY 3 4 1 ($2.30) 647 -7.0%
TWO 7 1 0 $21.80 655 67.1%
Average -12.8%
Click to enlarge

Here's the first test: You buy the security at the closing price of the day before the Ex-Dividend date, and sell it at the closing price on the day after the Ex-Dividend date, thus eventually collecting the dividend. Note that in most of these cases, the actual payment date is about a month after the Ex-Dividend date so for the moment, this calculation ignores the time value of that money. Also, I've ignored the transaction costs, namely the commission your broker will charge for the round trip trade.

In the columns above, a "success" is if the sum of the dividend and the post-trade stock price is greater than the pre-trade stock price. The overall success ratio for this type of trade is uncomfortably close to 50%. Note that in one of the cases, TWO, this strategy would have been successful 7 out of the 8 times, but for the rest of the securities, the overall strategy would have failed much more often, and even the successes would have been relatively small compared to the failures.

For the purposes of this analysis, the holding period is defined as the period just before the first 2010 dividend, or earliest dividend available for the funds that have not been around that long, so as to be able to calculate the annualized rate of return. We will also simplify the calculation by ignoring the time value of the money that is sitting in your online account for the 87 days between investments, or 30 days in the case of ARR which has foiled the strategy by paying their dividend monthly.

So, with the exception of TWO the strategy of buying the security immediately before the Ex-Dividend date would have been a loser, which in a way comforts us that the fundamental rules of finance are still mainly in place.

Secondly, the idea was posed of the opposite strategy: Buy the security immediately after the dividend, when the price was beaten down, and sell it at the precise moment of its theoretical highest value, which is immediately before the Ex-Dividend date:

Buy the day after the XD Rate, Sell the day before (not collecting the dividends)
Symbol Success Fail Avg % Gain Total Gain/Share Annualized ROR
AGNC 7 0 6.60% $12.58 26.55%
ARR 5 3 1.30% $0.72 12.36%
CIM 4 2 -1.30% ($0.27) -3.73%
CYS 6 1 6.1% $5.48 24.02%
DX 4 3 3.60% $2.04 12.63%
IVR 4 3 2.25% $2.42 5.92%
MTGE 0.00%
NLY 6 1 2.94% $3.55 10.77%
TWO 3 4 0.90% $0.85 2.62%
Average 10.13%
Click to enlarge

Since MTGE has not been around long enough for very many dividend cycles I have omitted the results for that security, but 7 of the remaining 8 would have shown a positive return for the period, even though the strategy is far from foolproof. Note that in the case of AGNC the strategy would have worked all 7 times, and also note that for TWO, the success in the strategy of playing the dividend date would have been punished in the opposite strategy, which is riskier because your money is tied up for a longer time period.

Finally, there is the simple strategy of buying the security before one of the dividend cycles, and sitting on it for the entire period:

Buy and Hold
Symbol Initial Final Dividends Net Anualized ROR
AGNC 26.77 28.01 $11.20 $13.24 27.9%
ARR 7.33 7.64 $1.05 $1.36 23.4%
CIM 4.06 2.51 $1.20 ($0.35) -4.8%
CYS 13.05 13.22 $4.60 $4.77 20.9%
DX 9.03 9.18 $2.07 $2.22 13.7%
IVR 23.16 14.11 $5.94 ($3.11) -7.6%
MTGE 17.64 18.27 $0.10 $0.73 15.1%
NLY 18.6 15.96 $5.09 $2.45 7.4%
TWO 18.11 16.11 $3.08 $1.08 3.3%
Average 11.04%
Click to enlarge

Note that the ROR calculation was adjusted for the holding period per the first graph.

You can see that this strategy is not foolproof either, two of the nine, Invesco and Chimera, were net losers over and above the dividend because of general dividend shrinkage, especially in the second half of 2011. The rate of return for AGNC, which is well thought of in this industry, was roughly equally high in both cases, and the "skip the dividend" strategy would have been slightly more successful in the two worst performing securities.

For the most part, simple is good, particularly since you still have to subtract transaction costs from the second calculation.

So, what are we to make of all of this?

First of all, our faith in the fundamentals of finance is restored, in that for the most part, someone who defers consumption and accepts risk is entiteled to some reward, at least in this case.

Secondly, there are exceptions to most rules, TWO in the first case being the most notable. Some investigation is in order here. TWO has the highest percentage of institutional investors of the funds in our little universe at about 60%.

Thirdly, an excellently performing fund like AGNC will behave roughly the same whether you omit the dividend and collect the capital appreciation or whether you chose to make money the easy way, except that the tax treatment may be quite different. Keep that in mind in any of these cases.

Fourthly, simple is good. Maybe that should have been #1.

Keep in mind that, as we are so fond of saying, the world is chaotic and there are no guarantees on anything, and also, past performance is no guarantee of future results, which in the case of IVR and CIM might be a good thing.

Disclosure: I am long CIM, CYS, AGNC, TWO, IVR. I went long on the above just before the first 2011 dividend and am following the easy third strategy.I also still have a limit order to buy ARR on a pullback.