onthemove wrote:Itsallaguess wrote:I don't think anyone should place any prominence at all on dividend payment dates - after all, we've no control over them, and companies often change the dates and payout configurations themselves, so why should they seriously matter?
I'd say there are two arguments in favour of smoothing out dividend payments, if reasonable and possible..
Firstly...
There is an opportunity cost in holding cash
I'm aware that at the minute that I have rather a substantial portion of my portfolio in cash (unrelated to expenditure; completely down to where to invest it), and I keep reminding myself that as it is, I'm foregoing probably somewhere in the region of £300 per month in dividend income as a result. I really should get a bit more pro-active!
Take the extreme example, if all your shares paid out on the same day of the year, then you'd have to effectively hold your annual income as cash for (on average) 6 months.
But in practice, they
don't all pay out on the same day of the year.
I've had a look at the monthly breakdown of the income from the taxable part of my main HYP for the last two completed tax years - I've chosen to do it on that part only because it's the dividend data I have most easily available. I should probably add that that part is most of my HYP and I'm only aware of one bias in terms of which shares I put into my ISAs and SIPP - a historical bias in favour of putting REITs into them, due to the Income Tax savings being higher. The 2008/9 financial crisis convinced me that that was a mistake: although property was hit less badly by it than banks, it persuaded me that I wanted to sector-diversify the dividend income more than I wanted to extract the last drops of tax savings on it. That lesson took a few years to sink in, but for over 5 years now I've been redressing the balance by directing dividend reinvestments and new subscription investments in the ISAs and SIPP away from REITs - but there's still some years to go before their high weighting in tax shelters is fully eroded away.
I should probably also add that the months I'm using here are what I might call 'tax months', running from the 6th of the month to the 5th of the following month. There's nothing essential about this - it's just to fit in cleanly with the records the data are being taken from are tax records and therefore organised by tax year. Similarly, I've only done it for the last two completed tax years because that avoids messing around with the old notional tax credits system.
So here are the figures:
Month 2016/2017 2017/2018
April 5.6% 5.5%
May 10.3% 9.7%
June 14.4% 12.5%
July 14.9% 13.6%
August 6.4% 7.6%
September 15.3% 15.5%
October 4.0% 4.3%
November 5.1% 10.4%
December 5.6% 4.4%
January 7.5% 8.6%
February 4.0% 0.7%
March 6.9% 7.2%
By the way, note that quite significant shifts do happen from year to year, often for quite trivial reasons. For example, BT's interim payout shifted back a day from February 6th, 2017 to February 5th, 2018, a type of shift that happens quite often because many companies stick to paying on the same day of the week each year to avoid running into weekends and bank holidays. But that shifted it from 'tax February' to 'tax January' for the purposes of my analysis, which was enough to cause a decrease in the February figure of 1.1% of the year's income and a corresponding increase in the January figure. That is of course dependent on my exact choice of where to place the month boundaries - but no matter where you place them, some dividends are likely to shift around because of mere 1- or 2-day shifts in the exact payment dates.
I'm not saying that all the changes have reasons that trivial - for instance, the rest of the February change is due to the merger of Standard Life and Aberdeen Asset Management (which paid in June and January/February) to form Standard Life Aberdeen (which pays in September/October and May). And the big increase in November is mainly due to top-ups and a new purchase happening to include quite a few November payers (I say "happening to" because I pay absolutely no attention to payment dates when selecting purchases myself).
Anyway, back to my main point: one can work out from those figures roughly how one can expect a buffer account to behave over the year. E.g. if one reckons on taking 1/12th = 8.3% income at the start of each month from it, and one uses the 2016/2017 figures, assuming a balance of S% of the yearly income at the start, it goes something like:
Date Changes Balance
5 April S
6 April -8.3% S-8.3%
5 May +5.6% S-2.7%
6 May -8.3% S-11.0%
5 June +10.3% S-0.7%
6 June -8.3% S-9.0%
5 July +14.4% S+5.4%
6 July -8.3% S-2.9%
5 August +14.9% S+12.0%
6 August -8.3% S+3.7%
5 September +6.4% S+10.1%
6 September -8.3% S+1.8%
5 October +15.3% S+17.1%
6 October -8.3% S+8.8%
5 November +4.0% S+12.8%
6 November -8.3% S+4.5%
5 December +5.1% S+9.6%
6 December -8.3% S+1.3%
5 January +5.6% S+6.9%
6 January -8.3% S-1.4%
5 February +7.5% S+6.1%
6 February -8.3% S-2.2%
5 March +4.0% S+1.8%
6 March -8.3% S-6.5%
5 April +6.9% S+0.4%
(The balance has increased by 0.4% of the year's income over the course of the year, but that's simply because 8.3% is slightly less than 1/12th - it's merely the best 1-decimal-place approximation to 1/12th, which is easily accurate enough for this purpose!)
One doesn't want the buffer account to go overdrawn, so one should make S be at least big enough to make all entries in the Balance column positive, i.e. at least 11.0% of a year's income. If one makes S=11.0%, then the balance varies between a low of 0% of a year's dividend income just after the May 6th payment is made to the current account, to a high of 27.1% of a year's dividend income just before the October 6th payment is made to it, from which one might estimate that the average balance over the year is about 13-14% of a year's dividend income. If one guesses that the lost returns from holding that much cash are around 8%, that's very roughly 1% of a year's dividend income lost per year - and if that were £300 per month, it would imply that one's annual dividend income were about 100 * 12 * £300 = £360k...
In reality, I wouldn't cut it that fine, and would make S a month or two's worth of dividend income greater than that - say 25.0% of annual dividend income - to allow for some variation of the payment schedule from year to year. That would result in the buffer account balance varying between about 14% and 42%, for an average cash balance of about 28% of annual dividend income. The income lost to keeping that cash balance would still only be around 2% of annual dividend income, so one would still need annual dividend income to be about £180k to be losing £300/month of income from it.
Also, as you've said, in reality one's expenditure won't be constant from month to month - but that's an issue faced by anyone who lives on a regular flow of income, such as that from a normal salary, so it's not really a HYP-specific problem. But predictable extra expenditure can easily be catered for with a buffer account - one can easily cater within the above framework for e.g. the normal monthly transfer of income to be 8% of the yearly total, but with an extra yearly payment in December of 4% of the yearly total to cater for Christmas. Unpredictable extra expenditure does however require the normal 'emergency fund' reserves that everyone needs, HYPer or not...
A more serious issue is that of dividend cuts. It will be quite normal to experience one or two a year from a reasonably large HYP, and usually the increases by the majority of the HYP's companies will make up for them, but they can come in bigger spates, either because of something like an economic recession or just pure bad luck. A moderately big spate of dividend cuts could fairly easily cost you 10-20% of a well-diversified HYP's annual dividend income (something like the 2008-2010 events could cost quite a bit more). That's far more serious than the above cost of a percent or two of annual dividend income - both because it's far bigger and because it's far less predictable and so cannot be properly planned for, but only dealt with by contingency plans. So basically, your planning should include ways of dealing with potential dividend cuts, and that planning ought to be able to cope with income shortfalls far bigger than the bit that goes missing because of holding (almost) unproductive cash!
And the supply of potential good payers for some periods of the year is very limited, because most companies have financial year ends that are December 31st, March 31st or dates reasonably close to those, and most companies making half-yearly payments with the final significantly bigger than the interim (those two combine to make late spring and summer the peak period for the amount of dividends received - which is the main reason why my HYP's payments have such a concentration in May-September compared with October-April). One can doubtless find a company or two that fit the bill of paying in almost any month you want, but it will be a lot more limited in some than in others - and if you take an even slightly higher risk of dividend cuts or accept an even slightly lower yield in order to get the month you want, it could easily cost you more than the bit that goes missing because of holding cash...
onthemove wrote:Secondly...
Going back to the extreme example, if all shares paid at the same time, once per year, you'd spend almost 12 months with no real feedback on how things are going.
Well, actually it's if all shares
declared their dividends at the same time, once per year, that you would be left with no real feedback, and even that assumes that you're not paying any attention to company news other than definitive information about dividend payments. But again, that extreme example simply doesn't happen in practice.
onthemove wrote:Finally...
If you've got a choice of two candidate investments, otherwise identical, but one pays out in a month where you've got a lull in payments, personally, I'd make the payment date the deciding factor between the two, for the above mentioned reasons.
I suppose that in theory, so would I. But I can think of plenty of other tiebreakers I would use in preference to it, and the chances that all of them would turn out not to break the tie are miniscule!
Gengulphus