Breelander wrote:Gengulphus wrote: I haven't yet confirmed just how the FT actuaries figure is calculated...
Section 4, the formula is in section 4.1.3
https://www.ftse.com/products/downloads ... o_Calc.pdf
Thanks - a document I'd previously encountered, but the reminder of where to find it is definitely helpful!
The bits that are important here are actually in sections 4.1.1 and 4.1.2, since the point I was making was about the source for the dividend values used in the formula, not the formula itself. The formula itself is of some interest, though, because a detailed look at it and the description of its inputs reveals that it isn't exactly a description of a market-cap-weighted average, but more accurately a description of a market-cap-adjusted-for-free-float-weighted average. If dividenddata is using a market-cap-weighted average, that difference in the weighting of the average could account (or partially account) for some small discrepancies between the FTSE100 index values calculated by dividenddata and by the FT actuaries.
But quoting sections 4.1.1 and 4.1.3, with the crucial bits emboldened by me:
4.1.1 Dividend yield is a widely used measure of the income return on a stock or index. In calculating index dividend yields, FTSE applies a free float adjustment to the number of shares in issue for each constituent. The dividend used to calculate the yield is the dividend declared by the companies, i.e with no allowance for any tax credit.
4.1.2 Dividends for the previous 12 months are included in the dividend yield calculation from the XD date. If the dividend is declared in a currency other than sterling, the 4pm WM/Reuters Closing Spot Rate on the day before the dividend is XD is used to convert the dividend into sterling unless the company announces a sterling equivalent which is used instead.
So each declared dividend enters the calculation when it goes ex-dividend, and leaves it again 12 months later (*). The dividenddata calculation is described in the sidebar of https://www.dividenddata.co.uk/dividend ... et=ftse100 as "This is based on the current share price and the total dividends declared in the previous 12 month period. In general, the dividend declaration date is used as the cut off date", which says that for it, a similar rule applies but with the critical date being when the dividend is declared, not when it goes ex-dividend. And I've confirmed that this really is the case - Itsallaguess's December dividends post allows one to find some FTSE100 companies that have a declared-but-not-yet-ex-dividend dividend and one can then check the details - for instance, it shows Royal Mail as having a dividend that goes ex dividend December 6th (this Thursday), https://www.dividenddata.co.uk/dividend ... et=ftse100 shows Royal Mail as having a yield of 7.70%, https://www.dividenddata.co.uk/dividend ... y?epic=RMG shows that 7.70% is the result of dividing a dividend total of 24.3p that includes that dividend by a price of 315.4p, https://www.dividenddata.co.uk/dividend ... y?epic=RMG shows that if one instead takes dividends that have gone ex dividend in the last 12 months, the dividend total is a bit lower at 24.0p and so the calculated yield using the same share price is also a bit lower at 7.61%. I can't confirm that that those are the figures the FT actuaries use with the same sort of detailed check of their calculations, but they are what the document says they use and that's probably as good as we're going to get...
So basically, it looks as if the dividenddata and FT actuaries figures are calculated from different dividend figures when the company has a declared dividend that hasn't yet gone ex-dividend and that dividend hasn't been held compared with the previous year's equivalent payment, and quite possibly the weightings of the averages they use are subtly different as well. So no real surprise that they result in slightly different FTSE100 yield figures, once one has dug into the matter deeply enough!
(*) As an aside, I do wonder whether that's to be taken absolutely literally. The reason is that the ex-dividend dates for a dividend and the equivalent dividend in the following year are seldom exactly 12 months apart, because they're both typically on Thursdays due to LSE rules / strong guidelines. That means that they are most commonly 364 days apart, causing them to drift towards earlier dates in the calendar by one or two days per year, next most commonly 371 days apart once every few years to correct for that drift, with other gaps less likely but possible if a company wants to materially alter its payment schedule. So it is not uncommon for say a final dividend to go ex-dividend and there to be a day or two when both it and the previous year's final have gone ex-dividend in the 12-month period, causing a spike in the company's contribution to the FTSE100 yield figure calculated... I'm a bit uncertain whether the actual calculation performed uses a not-quite-so-obvious interpretation of "the previous 12 months" to adjust for this, or just lets the spikes get into the final figure because they're too small to worry about - but it probably doesn't matter much!