Measuring your return (again)
Posted on October 12, 2007 by Richard Beddard
Filed Under Investing |
There’s more than one way to crack a nut. And deciding the right method of performance measurement is proving a hard nut to crack.
Graeme’s shared a spreadsheet showing how he calculates time-weighted performance.
If you’re new to this discussion, it follows Monday’s post, How are you doing? In which I explained the money-weighted method I use to calculate the return on my portfolio. It’s easy to calculate because the XIRR spreadsheet function does most of the work for you. But there’s a complication. The return is influenced by the money you pay into or take out of your portfolio during the period.
The answer is time-weighting if, like professional money managers, you want to remove the influence of cash flows. Graeme tells me his spreadsheet is:
…very similar to the way in which indices are calculated, rebasing after each change. In the case of an index the change is the addition or deletion of constituents. In this case it is the addition or removal of cash.The crucial thing about the time weighted method is that it assumes that you want to neutralise the effect of the inflows and outflows of funds. This ideal for measuring the performance of a fund manager who has no control over whether people decide to invest in a fund or take their money out.
It is less clear that it is always ideal for private investors. As you said, it may ignore timing decisions. One solution is to consider all your savings to be a single portfolio, although choosing an appropriate benchmark may be tricky.
You can see how it’s done in this version on Google documents, I’ve put the cell formulae in red, as close as I can to the cells they refer to.
It’s really not that tricky, once you have the spreadsheet set up. The only complication is you need to revalue your portfolio every time you pay money into it, or take money out of it.
As for which is the better measure for the private investor, I have to admit, I’m flummoxed. What I ultimately want to know is how much money I’ve made, and for that purpose XIRR seems adequate. Helpfully, for committed XIRR’ers Graeme has some good news. One criticism he makes of Internal Rate of Return calculations like XIRR is that there can be more than one right answer. But he’s spent some time looking for examples, and he tells me:
…it is extremely unlikely for a long only ungeared portfolio of bonds and equities. However for portfolios that are geared, high risk, or contain derivatives (i.e. anything that might cause very high returns or losses) it becomes likely enough to worry about.
However Graeme’s spreadsheet also demonstrates the problem with XIRR. The annualised return using XIRR for the model portfolio in his spreadsheet is under 9%. Using the time-weighted method it’s over 15%. Looking at the cash flows it’s clear that the portfolio had less money in it when it was doing well, and more in it when the value of the investments were falling. That depressed the money-weighted (XIRR) return, which perhaps makes it less comparable to benchmarks like the FTSE All-Share.
Although I’m intrigued by this suggestion…
Comments
3 Responses to “Measuring your return (again)”
Leave a Reply
Hi Richard,
You have done it again, another great article. Your two articles on measuring returns have been really useful to me so many thanks.
After thinking about it, I think you should use both measurements.
From my last post you can see I use a time-weighted measuring method because I truly want to discount the money I contribute on a regular basis. My regular contributions have no impact on my investment decisions at all.
However, in order to compare this to what a bank would have to pay me to get the same return then the XIRR is a perfect sanity check.
As I hoped, my XIRR is slightly higher than my time weighted return over the same period … phew.
PS I did not understand the post from ‘Windy’ so perhaps you or he could explain it with an example, if either of you have time.
Thanks Robin. That’s music to my ears
Although, this one was down to Graeme sharing his spreadsheet and explaining it to me.
Windy’s idea (also suggested to me by Le Fras, who helped me with the first post) is that if you use money-weighting you can make the FTSE-All Share, say, comparable with your portfolio rather than the other way round (as in the case of time-weighting - where you exclude the cash flows from your return calculation).
To do that you’d set up a dummy portfolio alongside your own, putting the same funds into and withdrawing the same funds out of it, at the same time. Then you’d calculate the XIRR on the dummy portfolio and use that as a benchmark.
Thanks, that is clear now and a clever idea.