How close do dividend futures come to matching the actual values for quarterly dividends per share that are reported for the S&P 500 by Standard & Poor?
We have two primary sources for dividend futures data for the S&P 500: the CBOE's four quarterly dividend futures options contracts (DVMR, DVJN, DVST, DVDE) that are set to expire in the months that mark the end of each calendar quarter and IndexArb's bottoms-up estimation of the dividends per share that will be paid out between the date that a given quarter's dividend futures contract will expire and the present.
We'll limit our comparison to each quarter since 2010-Q1, since that marks the time when the CBOE introduced its dividend futures options contracts. The chart below shows how each of our sources compares to the actual amount of dividends that were reported by S&P for each quarter since 2010-Q1.
Overall, we find that the CBOE's final projection of what a given future quarter's dividends per share will be ranges within 5.4% to the low side and 4.9% to the high side, with the overall average overstating what Standard & Poor reports the actual amount of quarterly dividends to be by 0.4%.
By contrast, IndexArb's final projection of what a future quarter's dividends per share will be runs from 7.1% to the low side and 4.9% to the high side, while averaging 1.9% to the low side from the values recorded by S&P. That discrepancy is primarily due to how IndexArb estimates dividend futures data, as the amount of dividends per share that will be paid out between the present and the end of the indicated quarter, where the quarterly dividends that will be paid for a given quarter is found by subtracting the value presented for the preceding quarter from the value presented for the quarter of interest.
Further complicating the issue, once the preceding quarter has ended, there is no good way to use IndexArb's dividend futures data to find out how the dividends that will be paid out by the end of the current quarter might be changing. Consequently, IndexArb's final projection of the amount of dividends that will be paid for a given quarter is made some three months earlier, before the dividend futures contract for the preceding quarter expires.
That means that IndexArb's final projection for a given quarter's dividends per share misses the changes that might occur for the current quarter's dividends once it becomes the current quarter - and really represents what investors expected three months earlier. Since most companies tend to increase their dividends over time, that's why IndexArb's final projection for future dividends runs to the low side of the actual value.
By contrast, one can find the quarterly dividends that will be paid out for the CBOE's dividend futures by dividing the presented value for the quarter by 10. It doesn't get much easier, and the CBOE's dividend futures for the current quarter run all the way to the end of the dividend futures contract for the quarter.
The downside to the CBOE's dividend futures is that it periodically experiences price anomalies - where occasionally low trading volumes for the dividend futures option contracts can cause some havoc with the indicated value of dividend futures. Typically, those kinds of anomalies last no more than a week, until options traders recognize the money-making opportunity associated with the discrepancies. The most recent anomaly occurred on Monday, 10 June 2014, affecting the DVMR, DVST and DVDE dividend futures contracts, and lasted just one day.
There are other issues that drive differences between the dividend futures and actual dividends. Probably the most significant is what we call "term mismatch", which is the situation where the dividend futures run according to the options contracts that expire on the third Friday of the month ending a quarter, while the data that S&P reports goes all the way through the final day of the calendar month ending the quarter. A recent example of how term mismatch works can affect the observed difference between dividend futures data and S&P's reported dividend data be seen in the data for 2013-Q4 and 2014-Q1.
The last issue we'll note is that the weighting of the components of the S&P 500 also contributes to the differences between dividend futures and actual dividends per share. This might be considered to be part of the term mismatch issue, but is something that can play a role given how and when S&P determines how the dividends paid by the index' component companies are weighted to produce their dividends per share figure.
Overall however, we find that the dividend futures data comes out to be remarkably close to the official values produced by Standard & Poor. And that's your quick guide to the data that defines what investors expect for the future in the stock market!
Previously on Political Calculations
- Whither Dividend Futures - we've discussed our data sources for dividend futures before, but we didn't measure how close they come to actual dividends!