Bree's HYPish Portfolio - Christmas Review 2016

[Gengulphus]

I've just noticed a distortion in that I had a buying splurge towards the end of this year, so my units have increased with few dividends being paid from them. This will drag down the inc per unit, naturally. In tandem, I calculate the income using descrete quarters which when totalled are quite different to the annual figure. That's only to be expected, but this year the difference is quite large - difference turns "growth" into "negative growth".

Well, there are distorting factors involved: seasonality and 'dividend lag'.

Seasonality is caused by the preponderance of companies with December 31st or March 31st year ends, or dates close to those, coupled with the tendency to weight dividend payments towards final dividends. Together, they tend to lead to more dividends being paid in late spring, summer and early autumn than in late autumn, winter and early spring. The only really good way of adjusting for seasonality that I'm aware of is simply always to work with a year's worth of dividends.

'Dividend lag' is caused by the delay from the ex-dividend date to the payment date, and results in there being a period of (typically) a month or two after you buy a share when there is no chance that it might pay you a dividend, and conversely, a corresponding period after you sell a share (assuming you do such a thing!) when there is a chance that it will yet pay you a dividend. Basically, it's just a delay in receiving cash you're entitled to, not any change to that entitlement - and the financial system is full of such delays, such as the couple of trading days from selling a share to actually receiving the cash when the sale is settled, the delay from becoming entitled to interest on an interest-bearing bank account (which is generally happening more-or-less continuously) to the account's next interest-payment date, or the typical one or two weeks from a takeover becoming definite to actually receiving the proceeds.

If desired, one can adjust for 'dividend lag' in determining how a portfolio is doing by accounting for dividends when they go ex-dividend rather than when they're paid: that becomes capable of affecting you immediately after you buy, and incapable of affecting you immediately after you sell. The trouble is that it's more work - you cannot straightforwardly read it off your broker's cash statement, and very occasionally a dividend is cancelled after going ex-dividend and so needs reversing in the accounting. (To give an idea how occasionally, I've experienced that just once in about 15 years of HYPing, generally with a 30+ share HYP - the occasion was BP cancelling its first quarterly dividend in 2010 after its Gulf of Mexico disaster.)

And the "If desired" at the start of that is important. Basically, it's something that is desirable when trying to measure the income-earning power of the HYP, as it removes a distorting factor in that measurement, but most definitely undesirable when making budgetting decisions about how/when to spend that income!

Gengulphus

[Arborbridge]

Gen,

When I give the end of year figure, I have always been consistent in calculating the income per unit in the way Bree does it: i.e. total cash received/total units at year end.

I also keep another set of figures which are calculated quarterly based on cash received/number of units at the beginning of the period. This is a crude attempt to approximately get around the dividend drag and create more points for a chart. One can then add the four quarters, which naturally produces a different number to the "raw" first method. This year, as I commented before, I believe latter method will give an increase in income per unit, whereas the first method produces a decrease in income per unit.

Whatever one does, it is rather like pondering "which type of yield" is one quoting. But as Terry once wrote, it doesn't much matter provided one is consistent, particularly for comparison purposes.

[JMN2]

Donkey's years ago in my previous life I used to be in charge of PnL for a multi-billion $ hedge fund, operations, middle-office, treasury, fx, hedging, repos, etc, in the morning reconcile the prime broker reports, post ex-dividends, reconcile cash, get the PnL's out early afternoon, get the daily trades booked, hedge fx bonds unrealised pnl work on NAV estimate liaise with PB finalise NAV grab lunch get home SWT shave eat sleep shower rinse repeat.

I won't unitise. Never had a single mistake (delays yes) with dividends or anything with my SIPP ISA HYP, have no clients, it's all just for myself so why make my life like it used to be, difficult, complicated. Ballpark is enough.

[OZYU]

My first post, hello there.

What an interesting thread.

I have always unitised since I started investing, since the days when we did it in our investing club, now long wound up.

I find it useful to unitise and it does not mean extra work once set up.

I have settled some time ago for preferring 12-months cumulated divis per income unit over the overall income divided by number of units, as I think it gives a fairer idea of what it happening as months roll by. I assemble this into monthly slots to reduce the size of the summary sheet and thus follow how the divis are increasing, net of re investing and new funds effects, continuously.

I also think that the yield calculated from these cumulated divis per unit is a fairer measure than the plain trailing yield. Certainly the differences are now smaller as my portfolio in very mature, but years ago at the intense portfolio building stages, the differences were larger between the two methods.

My results for 2016 in the ISA, which is HYP-like, are

Increase in cumulated 12-months trailing cumulated divis per income unit: 7.11%
Plain trailing yield 5.01%, yield from trailing cumul divis per unit: 5.22%
Increase in dividend income: 19.3%
Increase in portfolio value: 17.3%

I also calculate acc unit data as a natural by-product to benchmark against ftas tr.

Currently all income is re invested and there are fairly regular injections of funds, but this might very well change soon.

So a satisfactory year, no grumbles.

I fear that 2017 might not be as good, no particular reason, but that is how I feel about it, although I will stay broadly fully invested as usual anyway so as not to disturb the growing divi stream.

I also run a small trading mostly small/mid cap portfolio outside the ISA tax shelter. I have not unitised that one because I am very cavalier with it in all sorts of ways when capital in required for other causes.

Ozyu

[Gengulphus]

When I give the end of year figure, I have always been consistent in calculating the income per unit in the way Bree does it: i.e. total cash received/total units at year end.

Not the way I would do it! I would divide each dividend payment by the number of units at the time to produce an income-per-unit figure for that payment, then accumulate those income-per-unit figures. So if e.g. I had a dividend payment of £100 when my HYP consisted of 1000 units, then I added some capital which increased the number of units to 1100, then I received a dividend payment of £55, my overall income-per-unit figure would be £100/1000 + £55/1100 = 15p.

Dividing the total cash received by the units at the end of the period gives £155/1100 = 14.09p, which is too low because it's attributing the £100 payment to too many units. And dividing it by the number of units at the start of the period gives £155/1000 = 15.5p, which is too high because it's attributing the £55 payment to too few units.

Also, my method has the advantage that the divisor one uses for each dividend payment does not depend on which period boundaries one uses - which means e.g. that the income-per-unit figures one gets for each of the four periods that are quarters of a year add up to the same as the income-per-unit figure one gets for the period that is the entire year. One should of course be consistent about the date used - in particular, if one accounts for dividends when they are paid, accepting that the results will be affected by 'dividend lag', then use the number of units on the payment date; if on the other hand one adjusts for 'dividend lag' by accounting for dividends when they go ex-dividend, then use the number of units on the ex-dividend date.

My method does have the disadvantage of requiring more divisions, which would have been a real pain in the days before computers! But those days are long gone - it's now just a matter of making certain one's spreadsheets reflect the calculations one wants done, and then just letting them do their stuff...

I also keep another set of figures which are calculated quarterly based on cash received/number of units at the beginning of the period. This is a crude attempt to approximately get around the dividend drag and create more points for a chart. One can then add the four quarters, which naturally produces a different number to the "raw" first method. This year, as I commented before, I believe latter method will give an increase in income per unit, whereas the first method produces a decrease in income per unit.

Yes, assuming you're increasing the number of units as time goes by, because dividing by the lower number of units at the start of a period produces a higher quotient. Decreasing the number of units as time goes by (which would be a rather non-HYPish thing to do) would produce the opposite effects.

But there's no actual need to accept either an increase or a decrease - just use the number of units at the relevant time, not at the start or end of the period.

Whatever one does, it is rather like pondering "which type of yield" is one quoting. But as Terry once wrote, it doesn't much matter provided one is consistent, particularly for comparison purposes.

Indeed - but methods that say the income-per-unit figures for the four quarters of the year don't add up to the income-per-unit figure for the entire year don't exactly meet the "provided one is consistent" condition!

Gengulphus

[Dod1010]

My reaction to all of these calculations is, 'and if you do not like the result what are you going to do about it?' In other words why are you doing them? To compare results between portfolios- your own or others? For your own satisfaction (or frustration)? If you are using the result as an instrument to measure what you need to do and then take action as a result that seems to me to be a good use of time. Otherwise?

I do very little measuring but at year end and at the halfway point, I do take a look at the yield from my various holdings and the likelihood of that increasing over the next year or two. Of course I like to see the actual numbers but I already have a good idea because I am aware of the income throughout the year.

I cannot say that the capital does not matter but in a HYP it is does not get priority. In what I call my value portfolio I concentrate on capital and in that case the yield is almost immaterial in most cases.

Because of exchange rates, HSBC has been my stand out stock in income terms followed by Shell, particularly on the capital front. I do not hold Lloyds or a miner so have missed out on these recovery situations. Otherwise it has been no more than a fair to middling year. More than enough income for my needs but not a vintage year and I suppose that that reflects the economy as a whole. Inflation will be the watchword for 2017 I should think, and if interest rates rise, even modestly, that may cause a few flutters at least in the short term.

A Happy New Year to all.

Dod

[OZYU]

Gengulphus has describe in detail what I meant by cumulated divis per unit, it is exactly what I do as divis come in as this quantum of divis needs to be related to the actual units at the time.

But I also assemble these automatically into monthly slots by pivot to take a summary view since the date calculates a yy-mm field.

As he said, this does not mean any work in these days of spreadsheets once set up. I have not altered my spreadsheet itself apart from tiny tweaks on graphs for probably ten years.

So Dod1010, no need to worry that we spend any time on this beyond recording all portfolio activity as it occurs in the normal way. I would guess a few minutes a day, irregular of course, on average.

Happy new year to all

Ozyu

[Arborbridge]

Indeed - but methods that say the income-per-unit figures for the four quarters of the year don't add up to the income-per-unit figure for the entire year don't exactly meet the "provided one is consistent" condition!

Gengulphus

Why not? Each is self-consistent, and if you compare 2016 with 2015 etc on the same basis, then it that is consistent. That was Terry's point.

Your method is more exact, but too laboured. I'm not after accounting standard accuracy - just an indication whether I am winning or losing over a period of years.

Arb.

[idpickering]

Indeed - but methods that say the income-per-unit figures for the four quarters of the year don't add up to the income-per-unit figure for the entire year don't exactly meet the "provided one is consistent" condition!

Gengulphus

Why not? Each is self-consistent, and if you compare 2016 with 2015 etc on the same basis, then it that is consistent. That was Terry's point.

Your method is more exact, but too laboured. I'm not after accounting standard accuracy - just an indication whether I am winning or losing over a period of years.

Arb.

Have a virtual rec Arb!! Well said.

Ian.

[Dod1010]

So Dod1010, no need to worry that we spend any time on this beyond recording all portfolio activity as it occurs in the normal way. I would guess a few minutes a day, irregular of course, on average.

Yes but my question remains. What do you with the records? Do they help improve your portfolio?

Dod

[OZYU]

Your method is more exact, but too laboured

Not in any way a reflection on what Arb said, which I also find a consistent approach, but this type of 'aside' has often puzzled me over the years on occasionally scanning tmf as a lurker.

Many people will record portfolio activity such as divis coming in, buy/sells, corporate action, fees .... The basic data input will be the same for most.

Anything beyond that, such as any amount unitisation, irr, graphs, etc... just gets done as an output without lifting a finger, thus is not laboured at all at it is set up once only in the design of one's 'system'.

Ozyu

[OZYU]

So Dod1010, no need to worry that we spend any time on this beyond recording all portfolio activity as it occurs in the normal way. I would guess a few minutes a day, irregular of course, on average.

Yes but my question remains. What do you with the records? Do they help improve your portfolio?

Dod

A good and valid question, as usual from you.

First I enjoy doing it and it is little hassle, but mostly it gives me varied views on the overall 'health' of the portfolio which occasionally leads me to alter 'tack'(to put it in nautical terms, fellow sailors among posters might also see the link with ozyu).

Although retired, I am still at the building stage not yet taking income. Maybe my approach will slacken as years roll by and we start to use the income. In your own case, as I read it, you are just enjoying the income steam from a fully matured portfolio, whereas I am still watching over this 'unruly child' a little more closely.

Not important in the scheme of things.

Ozyu

[tjh290633]

Not the way I would do it! I would divide each dividend payment by the number of units at the time to produce an income-per-unit figure for that payment, then accumulate those income-per-unit figures.

Gengulphus

A good point. Easy for me to calculate, too.

That gives higher results (TJH is my original method, while GEN is the suggested mthod):

Year to    TJH    GEN  
31-Dec-10  15.91  16.20
31-Dec-11  18.45  18.67
31-Dec-12  20.83  21.62
31-Dec-13  23.67  24.08
31-Dec-14  28.18  28.73
31-Dec-15  24.12  24.55
30-Dec-16  25.50  26.11

TJH

[Gengulphus]

My reaction to all of these calculations is, 'and if you do not like the result what are you going to do about it?' In other words why are you doing them? To compare results between portfolios- your own or others? For your own satisfaction (or frustration)? If you are using the result as an instrument to measure what you need to do and then take action as a result that seems to me to be a good use of time. Otherwise?

And my reaction to that is that whether to do something and how to do it are two separate questions. Obviously if a HYPer's answer to the former is "No, don't do it", then they don't need the answer to the latter at all - but that doesn't invalidate answers to the latter, it just makes them irrelevant to that particular HYPer.

And generally speaking, if it's a question of how to do it, I know how and I can see reasonably likely reasons why someone might want to do it, I find it simpler and easier just to share my knowledge and leave each reader to decide the "whether" question for themselves! Questioning whether to do it is more for cases where I can't see reasons to do it, or only reasons that I think are unlikely to apply...

I do very little measuring but at year end and at the halfway point, I do take a look at the yield from my various holdings and the likelihood of that increasing over the next year or two. Of course I like to see the actual numbers but I already have a good idea because I am aware of the income throughout the year.

I cannot say that the capital does not matter but in a HYP it is does not get priority. ...

Yes, though the income-per-unit measurement we're discussing is an income measure rather than a capital one - specifically, one designed to disentangle income growth through dividend growth, from income growth through additional investment. It's most informative when the HYPer is adding substantial amount of extra capital - and at the opposite extreme, for HYP1 which has neither added nor removed capital since it was created, and so has had its number of income units remain unchanged throughout, it's a simple scaled version of the income record and so tells one nothing that the income record doesn't tell one anyway.

And yes, it's perfectly possible to do such stuff on the basis of a much more informally arrived-at "good idea", and indeed I mostly do precisely that myself - e.g. I've never unitised my main HYP because it would be a lot of work to get answers that I can get accurately enough for my purposes from such a "good idea". But I'm both highly competent at the sort of approximate mental arithmetic involved in arriving at that "good idea" and confident that it will be accurate enough for my purposes, and I know people do vary a great deal on such things: some are liable to end up with a "bad idea" rather than a "good idea"! Knowing one's own limitations is (IMHO) an essential part of investing, and so anyone who is likely to do that would be well-advised to either make do without having an idea about it at all, or to calculate it properly.

A Happy New Year to all.

Likewise!

Gengulphus

[Dod1010]

Thanks OZYU. Yes I am living off my dividends.

This is a long way from Bree's OP (sorry Bree) but for my own HYP I have a yield calculated on actual dividends received against year end value of 4.7% which is good enough for me. On an unchanged portfolio, I am up in cash terms a mere 3.1% caused I think by the loss of Amlin and the fact that Amlin had a big special in 2015 before its loss. That is despite all my individual shares except for Cobham at least maintaining their dividends, some like Shell and HSBC with quite big increases as a result of exchange rates.

That plus individual year end yields is about as much as I need to know for my HYP although it is always interesting to know what has happened to capital values.

Dod

[Gengulphus]

Indeed - but methods that say the income-per-unit figures for the four quarters of the year don't add up to the income-per-unit figure for the entire year don't exactly meet the "provided one is consistent" condition!

Why not? Each is self-consistent, and if you compare 2016 with 2015 etc on the same basis, then it that is consistent. ...

But they're not consistent with each other, so if you compare one with the other, it's not on the same basis. And you did compare one with the other...

And one other thing that is important in this context is that when I post, my intended audience includes the entire readership of this board, not just the person I'm replying to. I'm not saying that I think you need to care about the inconsistency between your per-year and per-quarter figures, or to change what you're doing in order to deal with it - that's for you to decide. Instead, I'm saying that it can be dealt with and how, for the benefit of any readers who do care about it.

Your method is more exact, but too laboured. I'm not after accounting standard accuracy - just an indication whether I am winning or losing over a period of years.

On the "too laboured" point, there are two different aspects. One is accumulating the income-per-unit figures for each dividend payment, and that's just a matter of whether you first add up the figures per period and then do the unitisation calculations, or first do the unitisation calculations and then add up the results. At least if one uses a spreadsheet, neither need be appreciably more work than the other - it's basically just a matter of setting things up to do the calculations in the right order (that's starting from scratch, of course - if one is starting from an existing spreadsheet, whichever it already does is clearly less work than changing to the other!).

The other aspect is whether one accounts for dividends when they are paid or when they went ex-dividend. On that, yes, the latter is liable to be more work, due to the required data collection being more difficult. That's the price that needs to be paid to adjust out the 'dividend lag' effect if that's what one wants to do - but I'm not saying that anyone has to want to do it!

Gengulphus

[Arborbridge]

Gengulphus:

And you did compare one with the other...

Oh Gosh, I did comment on that didn't I:) That's the trouble with thinking aloud. However, in my annual reports I believe I have only used the most commonly used method hereabouts, which is the inc per unit = (income for the whole year/number of units) as the end of the year. To that extent I've used a consistent method.

However in my spreadsheet, a while back I started wanting a little more detail for a chart of income, so I started keeping an income per unit calculation each month (until 2015, each quarter). In fact, this probably isn't so far out from what you've suggested, since my number of units does not alter significantly in most months now.
My feeling is that this is a good balance between effort and accuracy. I doubt there is any benefit in going further for a mature HYP. Yes, I can see that looking up the number of units for each dividend and accumulating the result would not be a tremendous effort, but in the past two or three years, I've been trying to reduce efffort, so I'm not inclined to increase it unless proved worthwhile, however slight the increase might be. From what you've said before about your own efforts, I'm sure you appreciate that too.

And as Dod implies: what is the point? Other than re-assuring myself that the HYP income is progresssing satisfactorily, none at all, and my level of accuracy is sufficient for that.

Further, consider that some units in that calculation are not even producing income yet, you come back to what kind of income per unit is one calculating anyway.

Sorry Bree, this is all rather highjacking discussion of your own HYP, but I hope it is of general interest.

Arb.

[bluedonkey]

I won't unitise. Never had a single mistake (delays yes) with dividends or anything with my SIPP ISA HYP, have no clients, it's all just for myself so why make my life like it used to be, difficult, complicated. Ballpark is enough.

I do so agree!

[MDW1954]

I've read all this with interest. I don't unitise, as I've never quite seen what it would give me -- in other words, as Dod asks, what would I do with the information? I've been tempted, yes, because I like numbers and spreadsheets and calculations. But it's always come back to that central objection.

I do keep an annual end-of-year spreadsheet, which I have just completed for 2016.

That said, if I were living entirely off my dividends (I'm not, and am still building), then I might devote more time to HYP management. Possibly.

MDW1954

[tjh290633]

As one of the instigators of unitisation, I started doing it to get a better idea of how I was doing, at a time when I was adding capital and withdrawing dividends on occasions. Previously I had done it by adjusting the value of the FTSE100 whenever I added capital, but that was not a reliable method.

TJH