Ernie Ankrim's March-April '92 FAJ "Risk-Adjusted Performance Attribution" article is one that I came across in my doctoral research. So far it's the only one I've found that addresses this topic (I reached out to Ernie in the hopes he'll pen one for The Journal of Performance Measurement(R)).
The fundamental question: should attribution be measured against risk-adjusted returns?
Traditional attribution models (e.g., Brinson Fachler) assume that the portfolio and benchmark have the same risk, and attribute the excess return to three effects (allocation, selection, and interaction). But what if the risks are different? Then, the excess return may be partly attributable to the risk delta. Ankrim's idea is that by looking at the risk-adjusted returns, we've eliminated the risk differences and look solely at the contributions from the manager's decisions.
This article is more than 21 years old. But who does risk-adjusted attribution? Should we rethink our approaches?
In The Spaulding Group's attribution class I sometimes make reference to this idea, and was pleased to see Ernie's article. Perhaps this is a topic worthy of more discussion; what do you think?
p.s., I was reminded that Andrew Kophamel wrote an article titled "Risk Adjusted Performance Attribution: A New Paradigm for Performance Analysis" for The Journal of Performance Measurement(R). I will have to re-read it!