Yesterday I discussed the “pricing” effect. Today I want to briefly touch on another “new” effect. By “new,” I don’t necessarily mean that it’s a recent introduction of an effect, but it’s new relative to the other standard effects that we often see reported on. I don’t recall seeing anything in writing on these effects before, so hope that this will prove helpful.
This second effect is the “trading” effect, which reflects the contribution to the excess return that results from trading activity during the period. This should not be confused with the transaction cost measurement, which is a whole separate science, so to speak, that measures such things as “market impact” and “value average weighted price” (VWAP, for short) where assess how efficient the firm is in executing trades. The latter should, in theory, be part of attribution analysis, too, but I’m not aware that this commonly done today.
By “trading” effect, we are taking into consideration the trades that take place during the period versus a situation where no trading was done. Here, we essentially are comparing the results from a holdings-based model (which ignores trades) with a transaction-based model (which incorporates them into the analysis).
The way I’ve heard to derive this effect is to use both a holdings and transaction-based model, and take the difference: the result is the trading effect.
One might ask what the value is of this effect. Arguably, it has little value as it (in my opinion) points out the difference between the two models, where one would expect that the transaction-based results would be superior. Is it worth the time and cost to derive these results? For what purpose are they being employed? I’m aware that some folks do, in fact, calculate them; just not sure if I’d recommend doing so.
To me, the greater value would be to tie your transaction cost measurement analysis into attribution. As noted above, I don’t believe this is common today, but should be pursued. Something to discuss further, no doubt.