My dividends per unit-TJH

[MDW1954]

Over the period 1989-2018, that's an average annual percentage increase in dividends per unit of just over 10%, if my maths is correct.

Pretty impressive.

MDW1954

[nmdhqbc]

Hope that is clear.

I thought I was ok, but now am confused, in your second table (including 2018) the rebased RPI figures have changed, but the rebased dividends haven't

Regards,
Darka

No reason why they should. The start point data is unchanged.

Ah, I think I know the reason. The RPI is based on the actual value, which was rebased to 100 in April 1997.

TJH

I'm not sure what this means. You say rebased RPI to 100 in 97 but not sure that's in the data or why that would be done. It's the 88 rebased RPI which is no longer 100 whereas the rebased dividend is 100. So they're starting at a different point now.

[tjh290633]

Here is the table, updated to 5th April last..

How does one calculate the number of units when one is adding to that number throughout the year?

I mean, to get Dividend / Unit I can easily calculate my total Dividend receipts over the past 12 months and then divide by the number of Units currently held, but is that the right way? Or should one take the balance of units at the start of the period?

Any explanation would be gratefully received


Ian

Good question. I always use the number of units at the end of the period, but I don't think there is a "right" way.

TJH

[Arborbridge]

The system I've preferred for the past year or two is to calculate the dividend per unit using the number of units in play on the day the dividend is paid. Resulting dividends are accumulated as they come in by adding to the previous total.

This run of recent dividends might make it clearer:



The div/unit column is the dividend divided by the number of units on the day. The right column is the accumulation, which starts at 4.905 which happens to be how much had accumulated to end September.


Arb.

[Arborbridge]

Over the period 1989-2018, that's an average annual percentage increase in dividends per unit of just over 10%, if my maths is correct.

Pretty impressive.

MDW1954

Maybe my turn to be slow, but isn't it: 31.73/2.87 = 11.05 in 29 years, making 8.6% compound. Still good enough!


Arb.

[OZYU]

Over the period 1989-2018, that's an average annual percentage increase in dividends per unit of just over 10%, if my maths is correct.

Pretty impressive.

MDW1954

Maybe my turn to be slow, but isn't it: 31.73/2.87 = 11.05 in 29 years, making 8.6% compound. Still good enough!


Arb.

It is possibly a bit less than that. Your calculation is correct, but the result is flattered by the current figure which is not in line at all with the past. If you plot the data, and use the slope of the least squares fit, you get a more reasonable estimate on how the underlying divis are progressing. I suspect that a divi attached to a consolidation has got into the last data item, thus flattering it vastly, such an event is of course not a unitisation event anyway. In my experience, and looking at my own divi per inc unit plots over a few decades on our various portfolios, the curves do not have such sharp changes, they are much smoother.

Ozyu

[Arborbridge]

]

It is possibly a bit less than that. Your calculation is correct, but the result is flattered by the current figure which is not in line at all with the past. If you plot the data, and use the slope of the least squares fit, you get a more reasonable estimate on how the underlying divis are progressing. I suspect that a divi attached to a consolidation has got into the last data item, thus flattering it vastly, such an event is of course not a unitisation event anyway. In my experience, and looking at my own divi per inc unit plots over a few decades on our various portfolios, the curves do not have such sharp changes, they are much smoother.

Ozyu

That last year change is certainly huge and as you point out is unlikely to be "normal". Taking the 28 years and excluding the last one give 7.99% compound.

[MDW1954]

Over the period 1989-2018, that's an average annual percentage increase in dividends per unit of just over 10%, if my maths is correct.

Pretty impressive.

MDW1954

Maybe my turn to be slow, but isn't it: 31.73/2.87 = 11.05 in 29 years, making 8.6% compound. Still good enough!


Arb.

It is possibly a bit less than that. Your calculation is correct, but the result is flattered by the current figure which is not in line at all with the past. If you plot the data, and use the slope of the least squares fit, you get a more reasonable estimate on how the underlying divis are progressing. I suspect that a divi attached to a consolidation has got into the last data item, thus flattering it vastly, such an event is of course not a unitisation event anyway. In my experience, and looking at my own divi per inc unit plots over a few decades on our various portfolios, the curves do not have such sharp changes, they are much smoother.

Ozyu

Let's face it, there are multiple ways to produce an "average" figure. My interest, actually, was in the variability of TJH's annual income percentage change figures, and so I pasted the data into a spreadsheet. The figure I cited was a simple arithmetic average of the year-on-year percentage change figures. A couple of phone calls then came in, and I had to go out for the evening, and so the variability investigation has had to wait.

Re: statistical methodology, two comments.

First, compounding is very reliant on the choice of start and end years, as Ozyu implied. A technique that includes more data points (ideally all of them) will be better than one that includes just two. Compounding also implies a degree of causality, namely that the outcome in year n+2 is dependent upon the outcome of year n+1, etc. With dividends, that ain't necessarily so.

Secondly, while I agree with Ozyu's use of least squares, I prefer logarithmic linear least squares for a percentage growth time series. On this occasion, I didn't bother to do this: on a quick-and-dirty basis, a rough average is good enough, at least for my purposes. Which takes me back to the beginning: TJH's data shows a lot of variability. How would our own HYPs perform in such circumstances? Not so differently, I suspect.

MDW1954

[Itsallaguess]

Here is the table, updated to 5th April last:

.           Ordinary   Rebased    RPI      Change     Change
Year to     Divs/unit  Divs/unit  Rebased  Divs/unit  RPI   

05-Apr-18       31.73   1,105.17   275.80     27.30%   1.92%

Hi Terry,

Coming back to this recent April 2018 data, and given the recent posts on how the 27.3% change in dividends per unit seems to skew some trends that we might be trying to look, are you able to give a little more detail behind what seems to be a large increase in that figure for the last full year please?

I think it would be interesting to see just where that change came from at a typical HYP portfolio level, if that's possible?

Was it due to a small number of companies rapidly increasing their payouts, or is there something else going on?

Cheers,

Itsallaguess

[tjh290633]

I've had a quick look at this. There was a big increase in dividend income, but it will take time to identify the sources. Also I see some discrepancies in my figures which should be self consistent.

I've tried to keep special dividends out of the picture, and hope that I have been successful.

The cricket has just started and I have to go to the dentist later. Maybe later this evening.

TJH

[tjh290633]

I thought I was ok, but now am confused, in your second table (including 2018) the rebased RPI figures have changed, but the rebased dividends haven't

Regards,
Darka

No reason why they should. The start point data is unchanged.

Ah, I think I know the reason. The RPI is based on the actual value, which was rebased to 100 in April 1997.

TJH

I'm not sure what this means. You say rebased RPI to 100 in 97 but not sure that's in the data or why that would be done. It's the 88 rebased RPI which is no longer 100 whereas the rebased dividend is 100. So they're starting at a different point now.

Sorry, that should be 1987, not 1997.

TJH

[nmdhqbc]

Sorry, that should be 1987, not 1997.

Ah, yes that corresponds more with the updated table. But why does RPI now get a 1 year head start? It downplays your dividend growth a touch.

[tjh290633]

Sorry, that should be 1987, not 1997.

Ah, yes that corresponds more with the updated table. But why does RPI now get a 1 year head start? It downplays your dividend growth a touch.

The first RPI figure quoted is that for April 1987, and the income is for the year to 5th April 1988. I suppose that one could take the October figure as the mid point of the year if one wished.

However, in answer to the question from Itsallaguess in viewtopic.php?p=172686#p172686 I have found an error in the spreadsheet for that particular line, which should be:

.          Ordinary   Rebased    RPI      Change     Change
Year to    Divs/unit  Divs/unit  Rebased  Divs/unit  RPI   
         
05-Apr-18      29.24   1,018.49   275.80     17.32%   1.92%

Apologies for the error, and thanks for leading me to it.

Looking back at 2017-18, the ordinary dividends increased by 18% over 2016-17. I am trying to identify the source of the changes, but it is easier to spot those shares which stopped dividends than those which received more, either because of topping up or dividend increases. It is very difficult to spot major changes, but I have looked at hatches, matches and dispatches, or new holdings, topped up holdings and sales, either partial or complete.

Total disposals in 2016-17 were: Rexam and Premier Farnell. New holdings were Rio Tinto, Carillion (smack on hand) and Legal and General. Trimmings were: Reckitt Benckiser, Rio Tinto and BAE Systems. Topped up were: Glaxo Smith Kline, Lloyds Banking Group, Taylor Wimpey, Marstons, Pearson, William Hill, Marks & Spencer, British Land, Aviva and BT Group,

For 2017-18 there were no disposals and no new holdings. Trimmed IMI and Segro. Topped up have been: SSE, Imperial Brands, Marstons, BT Group, Shell B, Pearson, William Hill, Glaxo Smith Kline, United Utilities, Segro and South32.

Some holdings have been topped up more than once. I can only attribute the increase to the reinvestment of trimmings and accumulated dividednd into higher yielding shares. Trimming is usually about 25% and topping up is about 20% but varies according to the value of the holding.

Hope that sheds some light.

TJH

[funduffer]

Hi tjh, I am late to this thread, but a couple of points for discussion:

I had forgotten RPI was over 12% in 1989! Wow, it seems a long time ago we had double digit RPI, but it was all the rage in 1970's and 1980's. I wonder if we will see those days again in our lifetime?

I agree it is difficult to isolate why your most recent increase in income per unit is higher than average, given the number of top-slices and top-ups you do. Maybe that is the answer though, you are forever ratcheting up the yield by top-slicing lower yielding shares and replacing with higher yielding top-ups. Provided there are aren't too many cutters, you are bound to ratchet up the income per unit, as presumably each top-slice/top-up does not change the number of units much, unless you are adding new money as well?

FD

[nmdhqbc]

The first RPI figure quoted is that for April 1987, and the income is for the year to 5th April 1988. I suppose that one could take the October figure as the mid point of the year if one wished.

To be honest I still don't get it. Surely whichever RPI figure is used in the first line it should be rebased to 100 to match the dividends rebase like it was in your first table. I though that was the whole point of rebasing the two columns to compare them from a level paying field. I'm probably being dumb but I just don't see it yet.

[tjh290633]

To be honest I still don't get it. Surely whichever RPI figure is used in the first line it should be rebased to 100 to match the dividends rebase like it was in your first table. I though that was the whole point of rebasing the two columns to compare them from a level paying field. I'm probably being dumb but I just don't see it yet.

Sorry, I see what you mean now. Somewhere the rebasing got lost along the way. The corrected table is:

.           Ordinary   Rebased    RPI      Change     Change
Year to     Divs/unit  Divs/unit  Rebased  Divs/unit  RPI   
05-Apr-88        2.87     100.00   100.00                   
05-Apr-89        2.75      95.71   112.28     -4.29%  12.28%
05-Apr-90        4.33     150.98   122.89     57.74%   9.45%
05-Apr-91        5.75     200.34   130.75     32.69%   6.39%
05-Apr-92        7.97     277.71   136.35     38.62%   4.28%
05-Apr-93        7.33     255.30   138.11     -8.07%   1.30%
05-Apr-94        6.65     231.50   141.65     -9.32%   2.56%
05-Apr-95        7.93     276.14   146.37     19.28%   3.33%
05-Apr-96        7.81     272.15   149.90     -1.44%   2.42%
05-Apr-97        8.90     310.02   153.54     13.92%   2.42%
05-Apr-98        9.35     325.60   159.72      5.02%   4.03%
05-Apr-99        8.91     310.18   162.28     -4.73%   1.60%
05-Apr-00       11.96     416.65   167.09     34.32%   2.97%
05-Apr-01       12.42     432.57   170.04      3.82%   1.76%
05-Apr-02       13.82     481.20   172.59     11.24%   1.50%
05-Apr-03       12.95     451.20   178.00     -6.24%   3.13%
05-Apr-04       12.48     434.56   182.42     -3.69%   2.48%
05-Apr-05       12.96     451.25   188.21      3.84%   3.18%
05-Apr-06       14.09     490.63   193.03      8.73%   2.56%
05-Apr-07       15.07     524.76   201.77      6.96%   4.53%
05-Apr-08       26.09     908.86   210.22     73.20%   4.19%
05-Apr-09       22.76     792.84   207.76    -12.77%  -1.17%
05-Apr-10       11.91     414.74   218.86    -47.69%   5.34%
05-Apr-11       16.71     582.13   230.26     40.36%   5.21%
05-Apr-12       18.79     654.57   238.21     12.44%   3.46%
05-Apr-13       20.89     727.68   245.09     11.17%   2.89%
05-Apr-14       21.48     748.29   250.29      2.83%   2.12%
05-Apr-15       22.40     780.31   253.44      4.28%   1.26%
05-Apr-16       22.77     793.06   256.78      1.63%   1.32%
05-Apr-17       24.93     868.15   265.82      9.47%   3.52%
05-Apr-18       29.24   1,018.49   270.92     17.32%   1.92%

TJH

[nmdhqbc]

Hoorah

[MDW1954]

Terry,

Could you kindly explain the resulting difference between your initial table and this revised one? I'm looking at the first couple of lines of each, and not seeing whatever it is that poster nmdhqbc sees. Two glasses of rioja won't have helped, but it's all a bit cryptic at the end of a long stressful day.

Unless you've subsequently edited the initial post (which I doubt), then I'm at a loss.

MDW1954

[Darka]

Unless you've subsequently edited the initial post (which I doubt), then I'm at a loss.

AFAIK.

The 1st table was correct but missed off the most recent value as of 05-Apr-18.

The 2nd table was supposed to be the same as the first but with the 05-Apr-18 value added, however the rebasing went a bit wrong here I think, so...

The 3rd table is now correct, I.e. same as the 1st one but with the 05-Apr-18 value added.

At least that's my understanding which of course may be wrong...

regards,
Darka

[tjh290633]

Darka is correct. I found a mistake in a formula which gave an incorrect result for div/unit for 05-Apr-2018. Also the rebased values for the RPI got lost, somewhere along the way, now reinstated.

This is one of the hazards of inserting rows, where formulae need to be spawned down to ensure that they are correct. Errors creep in all too easily.

TJH