Getting our attribution right

In last Friday's WSJ, Cameron McWhirter had an article titled "To Quote Thomas Jefferson, 'I Never Actually Said That,'" in which he refutes many quotes that are incorrectly attributed to our America's third President (and one of our Founding Fathers, author of the Declaration of Independence, the country's first Secretary of State, and much more). He cites Anna Berkes, a librarian at the Monticello Jefferson Library who speaks of the "rampant misattribution" of such misassignment of quotes.

In performance attribution analysis, we must also be on the lookout for misattribution; what's the point of doing the analysis if it's going to result in figures pointing to the wrong cause?

For example, on a few occasions I've found global managers who fail to measure the impact of currency movement. This means that the selection effect is including it, and therefore possibly crediting selection decisions when the currency movement is the correct source.

I've written about the problems that holdings-based models can cause, especially when there's a fair amount of trading occurring within the portfolio: not only can we expect to see large single-period residuals (as opposed to multiple-period, that can result from the absence of a sound linking method), but also the misassignment of the effects (more to follow on this point).

The integrity of the attribution analysis depends on the model's alignment with the investment process, as well as the completeness and accuracy of the model's formulae.

Misattribution can result in giving credit where credit isn't due, failing to see problems that may exist within the investment process, and misleading clients and prospects about the sources of return. They can also call into question the value of the performance team. None of these are good things.