The results you get for these attribution questions vary slightly depending on the assumptions you make, however the examiner has suggested to me that he is happy to accept any "reasonable" approach. What follows below is his preferred approach - although it actually differs slightly from his main solution on the Examiner's Report for the April 2006 question, as he calculated the actual weights slightly differently - i.e. just based on the opening fund values.
The actual returns are:
r(usa) = 24.49%
r(jaa) = 46.15%
r(eua) = -13.33%
r(asa) = 40.00%
r(caa) = 0.00%
These returns are calculated assuming that the cashflows are invested midway through the year and also the approximation (1 + i)^t = (1+ i*t).
The notional returns are:
r(usn) = 30.29%
r(jan) = 16.92%
r(eun) = -13.67%
r(asn) = 53.03%
r(can) = 0.00%
For example, the notional return for the USA (in £ terms) is calculated as:
(120/100) * (1.90/1.75) -1 = +30.29%
reflecting both the increase in the dollar value of the fund and the appreciation of the dollar.
The actual weights are:
w(usa) = 49%
w(jaa) = 19.5%
w(eua) = 21%
w(asa) = 10%
w(caa) = 0.50%
These weights are again calculated assuming that the cashflows are invested midway through the year - which is probably the most reasonable assumption to make unless the question suggests otherwise.
The US actual weight was therefore w(usa) = 245/500 = 49%.
Combining the actual weights and returns estimated in this way then gives:
r(aa) = sum of [actual weights x actual returns] = 22.20%
which is in fact "correct", in the sense that the fund grew by 22.2% from 500 at the start to 611 at the end, with no net cashflow at the halfway point.
The examiner's solution of 22.81% here is inconsistent as it is based on weights caluclated using opening fund values and returns based on mid-year cashflows.
Finally, the notional weights are:
w(usn) = 50%
w(jan) = 15%
w(eun) = 30%
w(asn) = 5%
w(can) = 0%
So, the overall returns are:
r(aa) = sum of [actual weights x actual returns] = 22.20%
r(an) = sum of [actual weights x notional returns] = 20.57%
r(nn) = sum of [notional weights x notional returns] = 16.23%
So, total stock selection profits = r(aa) - r(an) = +1.63%
total sector selection profits = r(an) - r(nn) = +4.34%
The stock selection profit contribution of the US is then found as:
= w(usa) x [r(usa) - r(usn)] = 0.49 x [24.49 - 30.29] = -2.84
Using the same approach for the other sectors then gives the stock selection contributions for each as: +5.70, +0.07, -1.30 and 0.00%, which sum to the total of +1.63%.
The sector selection profit contribution of the US is then found as:
= [w(usa) - w(usn)] x [r(usn) - r(nn)] = [0.49 - 0.50] x [30.29 - 16.23] = -0.14
Using the same approach for the other sectors then gives the sector selection contributions for each as: 0.03, 2.69, 1.84 and -0.08, which sum to the total of +4.34%.
In the April 2006 question, the currency contribution was found as part of the overall sector selection profit and it was found in an analogous way to that just described for splitting out the overall sector selection profit.
You first need to calculate the returns from investing purely in each currency, e.g. for US dollars this was:
r(us$) = 1.90/1.75 = +8.57%
which is positive as the US dollar appreciated over the year.
The corresponding figures for the other currencies are then: -5.00%, -6.67%, +8.57% & 0.00%. The weighted average currency return, based on the notional sector weights is then equal to:
r(cn) = sum of [notional weights x currency returns] = +1.96%.
The currency contribution of the US dollar to the overall sector selection profit was then found as:
= [w(usa) - w(usn)] x [r(us$) - r(cn)] = [0.49 - 0.50] x [8.57 - 1.96] = -0.07
The other currency contributions are then found in the same way and are -0.31, 0.78, 0.33 and -0.01.
However, I don't imagine that many people got this right, particularly as nothing like this had been asked before and it was not clear from the question that it was required.
Ignoring the currency element, the above approach has been exactly what has been required in each of the three ST5 performance attribution to date. I hope the above is reasonably clear - a complete solution is given in the ST5 ASET.
Last edited by a moderator: Apr 16, 2007