issue with global performance evaluation

Hi, I was reading volume 6 of the CFA institute material. On pages 196 and 206, two different ways have been used for calculating the market allocation contribution. Can someone explain why has it been done so?

…and security selection… I’ve been looking for at least a sentence to explain this stuff to me but no show… I guess this is yet another formula for the crams…

Hey Loke I take that back. It seems we’ve both been switching the variables i.e using benchmark weight instead of portfolio weight to calculate security selection and using portfolio returns instead of benchmark returns to calculate the market allocation. The formula remains the same for the GPA and multiperiod analysis…

Schweser actually mentions this “discrepancy” in the curriculum as CFAI textbooks use 2 different formulas for market allocation (one in Micro attribution and another in Global Attribution). They even asked the institute about it, where they replied that both formulas measure the same thing… Actually, as you already precised, both formula yield the same result on an agregate basis, although the individual country results are different… This is why you got the 2% right but the individual allocations were different. Schweser concludes by stating that since we don’t have to perform those kind of calculations on the exam, we will not be troubled by this discrepancy. I personnaly doubt we won’t have to do those calculations, so better be prepared… PS: I already posted this in another thread relating to the same issue.

Thanks for that clarification CFALEB. I actually tried with the first formula initially and saw that it worked in period one. When I tried it in period 2 I got a rude shock, but I later realized I was the one mixing up the variables. Yeah, I agree we shouldn’t let the Schweser advise fool us…remember TB in the '07…

On a related note, what happened to the interaction term in reading 48? It shows up in reading 47 (micro-attribution), but not in reading 48 (global performance evaluation). Can anyone explain why they dropped it from the equation in reading 48?

Figured it out - interaction is combined into selection

here’s how i understand it… the “interaction” return is the same concept as the “cross-product” return - just a different name for it. whenever you have 2 or more components affecting returns (eg any combination of currency, stock-selection, sector weightings, asset class weighitings, etc) the impacts aren’t additive, they are multiplicative. So when you try to break down the total return into the component parts, you always get this “cross-product” term(s) as the balancing item(s) to account for the total combined impacts of the components. so, going back to page 196 the calc just says that the currency gain/loss is NOT just the change in FC against the DC because that misses out the cross-product. Eg. if a US investor is investing in a UK asset. If UK (FC) local asset gains 10% in FC terms, and if the UK currency (FC) gains 5% against the USD (DC), then the currency return is NOT just the 5% FC gain. The currency return is the Total DC return less the FC (UK asset) return. Total DC return = (1+10%)*(1+5%) - 1 = 15.5% So the currency return = (15.5% - 10%) = 5.5%. ie pure currency gain (5%) PLUS the cross-product of 0.5% = 15.5% so far, so good. So moving on to the currency attribution where you have several different local FC asset sectors or classes and several different currencies, the components of return are: 1) pure stock selection impact = impact of pure stock-picking in the local FC asset classes (ignoring the impact of allocation weights, and ignoring the impact of currency weights) = weighted av of the local currency FC asset Active returns in the FC asset sectors + 2) Market allocation impact = impact of over/under-weighting the different FC asset sectors, (ignoring the stock-picking impact within the sectors, and ignoring the currency weighting impact) = (w/av Portf FC return - w/av BM FC return) less the pure Stock selection impact (to isolate the cross-product) + 3) Currency allocation impact = impact of over/under-weighting the different currencies, (ignoring the impact of sector weights and ignoring stock-picking in each sector) = Portf Currency contribution (which includes the Portfs cross-product) less the Benchmark currency contrib (which also includes the BM cross-product). [These currency contribution calcs are the same as the first one above - ie (15.5% - 10%) = 5.5% so includes the 0.5% cross-product] These 3 added up give the total active Portf return in DC terms This is the same conceptual approach used in Micro attribution in reading 47 without the currencies thrown in for good measure. ie in the Domestic context, the components of manager portfolio return are: 1) pure stock selection impact within each sector, ignoring sector weight impacts 2) + pure sector allocation impact, ignoring stock-picking within sectors 3) + the cross-product return (called “interaction” between the two components) hope this helps…it looks better when I use numbers…but I had to follow the concept of the cross-products (interactions) so I made sense of it… BTW this cross-product also pops up in Reading 41a when trying to hedge the FC asset - the hedge is never perfect because the cross-product is un-hedged. And again in Reading 42g