Wildly - I’ll e-mail them. Edit: BTW - The way you should immediately know it is untrue is in the case where r= 0. If X and Y have correlation 0, does that mean that X has correlation 0 with everything that has a non-zero correlation with Y? So X is the price of corn and Y is the S&P 500 - Kellogg’s(say they have a correlation of 0) and Z is Kellogg’s which makes only corn flakes. Do X and Z have 0 correlation?
E-mailed. Edit: And BTW - since I’m typing here a lot - the above stuff about CAPM and least squares seems pretty confusing because CAPM says nothing about least-squares.
I’m pretty amazed that not only is the question wrong (correlation clearly not being transitive), but someone (with a CFA charter no less) actively came back and said that cov(a,b)*cov(a,c) = cov(b,c). That’s a Stalla AVOID note then.
I still like Stalla and chrismaths should just see this as a sign of limitless opportunity.