Spin-offs and the Portfolio/Investment Overlay Graph

[faidas]

Hi Mark

How is the math of the Portfolio/Investment overlay (value
and price +distributions graphs) handling spin-offs?
My question arises from the following:

I own Altria (MO). On 3/31/08 MO spun-off Phillip Morris
International (PM). I handled that transaction as a spin off
(one share of PM issued per MO share owned).
Everything lin my portfolio looks ok pricewise.

At the closing of that day (3/31/08) the dollar value of my
portfolio was slightly up (~0.1%)for the day.
This was evident in the portfolio/inv. overlay value graph.
However the port./inv overlay price+distributions graph
(normalized to 100 on some initial date)
showed a drop of -1.53% for the day. On other days without
a spin-off the two graphs are in agreement i.e absolute
dollar value rise/fall and percent rise/fall.
I suspect this is the result of the spin-off but how?
Can I make the two graphs agree?

Thank you

Homer

[Mark]

Hi Homer,

A spin-off just records a Return Of Capital distribution in the parent, and an equivalent value new purchase in the child.

You can see the math used for this graph type from the documentation here:

http://www.fundmanagersoftware.com/help ... tovly.html

Basically, the portfolio price is a hypothetical price, based on a weighted average which is adjusted on a daily basis. Each day the % increase of each investment is calculated. This %increase includes distributions, so even though the parent share price dropped a lot, when you add back in the Return of Capital distribution this will compensate for that drop. The hypothetical portfolio price is calculated with a weighted average of each investment as of the beginning of the day. I'm guessing that your parent decreased that day (even including the distribution), but your new child increased. The daily weightings would only include the parent performance on that day, since you owned $0 in the child as of the beginning of the day, giving the child's performance weighting for that day a factor of 0.

I'll have to give this some more thought on ways to improve this, but I think the problem stems from not knowing when in the day the transferred money went to work for your portfolio. In this case, it went to work on the beginning of the day, but Fund Manager is essentially treating it as though it went to work at the end of the day. Changing this assumption may correct this case, but could cause similar, but opposite, artifacts when purchasing/transferring within or towards the end of the day. One solution would be to increase the resolution of the weighting adjustments, but Fund Manager doesn't track absolute time within a day for your transactions, so this would be a rather large change.

If you followed all of that, congratulations... If not, the bottom line is that this is due to an artifact in Fund Manager's algorithm for this graph type.

[faidas]

Thank you Mark

I think I follow you. Based on your formula the percent change of
the portfolio on a given day is the weighted average of the
changes of individual investments. But the weights are taken
from the day **before**. The day before the weight of the

child is zero and the parent is still "pregnant".
So the drop in the
parent is not counterbalanced by the "rise" of the child.
And it is hard to define the rise of the child %wise since it
starts from zero.
I guess you need to "renormalize" parent and child weights.
(This is similar to the problem faced by indices such as Dow
Jones when they have dividend payments spin-offs etc).

One way to do it, is on the day of the spin-off to treat parent
and child as one investment unit when calculating the average
to circumvent the zero weight of the child the day before.
On the next day the formula will treat them separately.
For example a portfolio has only one investment A on day 0
10 shares trading at $10
On day 1 A splits into P (10 shares trading at $5) and C
(10 shares trading at 5 also). There was zero change
but the current formula will show a 50% drop in day 1.
I suggest for day 1 to compare the price of A before the
spinoff with the sum of P+C as if they are one investment.
On day 2 P and C are separate.
May be there is another way to achieve the same result.
This may also correct the annualized yield shown on top of
individual investments that had spin-offs. My MO now
shows a -30% return which I know is not true.
Please look into this because a portfolio with many large
spin-offs will appear to underperform comparative indices
systematically.

Thank you
Homer

[Mark]

Hi Homer,

Thanks for the idea, that is an interesting way to look at it. I can see it helping in this case, but ideally I can figure out something that works in all cases where you acquire new shares. For example, even when you purchase out of pocket new shares, this would cause a similar problem.

On the return issue... Your yields should still be correct, as they do factor in distributions. If you check the recorded Return of Capital Distribution, does the distribution per share look correct, to account for the corresponding drop in share price? For example, if MO dropped from $75 to $25, you should have a $50/share distribution for flat performance. If for some reason your distribution is less than this, this would reflect poorly on the performance of this investment. Look in the Data Register with the "Distributions (all)" data type to see the per share distribution amount.

[faidas]

Mark
yes the yields should be correct but there is a twist to it too

In the spin-off window you specify the number of shares
received and the "cost basis". The cost basis serves also as the
value of the distribution of the spin-off. But it can also be the
cost basis allocation to the "child". This can make a huge
difference. For example I bought MO several years ago and
my cost basis was ~20. Before the spin-off MO was trading at
~72. After the spin-off MO was at ~20 and PM (the child) at
52. If you alocate the 20 cost basis proportionately 6 goes to
the parent (MO) and 14 to the child. So the distribution
appears small (only 14/share) but the drop upon the spin-off
huge (52/share). This creates the apparent negative return.
If you assign a cost basis of ~52 then distribution and price
drop are roughly equal but the cost basis has not been
allocated correctly to the parent/child shares.
May be another piece of information should be entered in
the spin-off window. Cost basis allocation and separately
initial post spin-off parent and child prices.

Homer

[Mark]

Hi Homer,

Excellent point. To be complete here, you have to specify both the cost basis and market value of the child shares. The parent would have 2 distributions:

1) A Return of Capital for the amount of the cost basis transferred to the child
2) A Distributed Distribution (not Return of Capital) for the difference between the market value and the cost basis

The child would have a Transfer In of shares, where the market value is different from the cost basis. (Currently, the 8.x versions use a Purchase transaction here, where the market value of the purchase is always the same as the cost basis.)

If we apply your example below, the child cost basis would be 14 and the market value would be 52. The parent would get a Return of Capital for 14 and another distribution for 38 (52 - 14), so the total of the 2 distributions would be 52, keeping the performance correct, while also keeping the cost basis correct. The child would get a Transfer In with a market value of 52 and a cost basis of 14. Yields are calculated using the market value of the transfer, so the performance and basis of the child would also both be correct.

I'll plan on adding a market value field into the spin-off dialog for 9.0. The 8.x versions do not have a Transfer In transaction. This has been added to 9.x though, so will be useful for this situation. Unfortunately, this means you cannot manually correct this situation in 8.x. I would suggest leaving the purchase at the cost basis value, and not recording the other distributed distribution in the parent. (Basically, leave things like they were recorded from the spin-off dialog.) Your overall portfolio yields will still be accurate, but the MO performance will be too low, and the PM performance too high (by the same amounts). When 9.0 comes out you can change how this is recorded, so both have their proper performance.

Thanks for pointing this out.