Well, perhaps coming from the background of science, I have a different of what "test" and "prove" mean than someone who doesn't, but I don't know that this makes these ideas inapplicable in this realm.
Understand that until not very long ago I hadn't really thought about valuation a lot beyond having this nagging sense that I should be finding some time to pay more attention. As each person would offer some idea on this forum, I would pay enough attention to make sure I understood the concept, but I have had this building sense of dissatisfaction. You may have noticed this cropping up in some earlier posts, e.g., the bit about "value" to the investor actually being difference in share price, not some underlying aspect of the company that supposedly related to share price and this being a further variance inducing factor to separate real current value from any computed figure based on financials.
As we started the current discussion, however, this jelled for me a bit more since I started to realize that however good the argument for such and such a calculation being a "true" measure of valuation, until one put it to the test, it was really no more than a hypothesis about a possible metric.
I think this is part of the reason that we have seen as many different metrics and different interpretations of those metrics as we have. Ultimately, these come down to "this way of computing makes me feel more comfortable" and "this way of interpreting the number feels right".
But, there is no reason to be satisfied with that level. To be sure, all of these metrics involve a certain amount of interpretation in order to come up with the base figures ... although that seems to me to be something that falls within the realm of a fairly well-defined art and the differences are ones that mostly could be noted and codified. And, to be sure, since the reason for computing these figures is to gain predictive advantage over the future (a familiar phrase in science) and there are a lot of unknowns operating, not unlike any problem in behavioral science, we don't expect the predictions to be exact. But, we certainly should be picking metrics based on which is most accurate with the least variation and where the "errors" in prediction can be the most satisfactorily explained. Why wouldn't we want that?
And, given that, if we had a proposed metric, why wouldn't we start out testing both the metric and the proposed rules against history to see how well it would have worked then? To be sure, doing well against history is no guarantee of the future, but if it can't explain the past, why would we think it would work in the future.
Were I not up to my neck in alligators with day job work, I would love to leap into working this out, but as it is I am grabbing odd moments here and there to offer thoughts. Perhaps this thread would prefer that there be a new thread to explore valuation in this way, but given that there is any kind of consensus on the thread that we want to know more about valuation in the context of G&K, I can't see why we would want to limit that exploration to valuation metrics which were only intuitive.
As you say, if nothing else, perhaps I can make some of the valuationists work a bit harder at explaining their opinion. |
