With customer-based valuation drivers, ROIC, and other related measures, it's generally wise to assume regression towards some sort of global average. If so, the billion dollar questions become (1) what the global averages are and (2) how fast the shrinkage will occur. [1/5]
It's striking how often this is ignored in valuation (by myself included!). But this is old news to any hierarchical modeler, and of course to folks like @mjmauboussin (and folks like Rappaport and Modigliani/Miller) who have more diligently thought about it/how to model it.[2/5]
The tough part is that to do shrinkage estimation for overall valuation in a truly Bayesian way requires bringing all the relevant companies into a single joint valuation model, and shrinking off the right things. I see this on narrow problems, not multiple DCF models. [3/5]
CLV and CAC figures evolve over time, and there is a clear (but nuanced) link between ROIC and CLV vs CAC. As with ROIC, big numbers attract competition that chips away at excess. [4/5]
