OK, I have been doing CFA stuff for a while, but I still flirt with CFA Heresy from time to time. I understand that there is this rule that says: “There is no premium for taking on diversifiable risk.” Only undiversifiable risk gets a risk premium. Now is this a financial principle, or is there some kind of mathematical proof that if diversifiable risk could be taken, that the world would explode or time travel would be possible or some other paradoxical thing? (OMG, it’s ARBITRAGE! 
) And if so, can we tell a priori if a risk is a diversifiable kind of risk or not. I might find it easier to accept that only diversifiable risk is paid if it meant that we might still discover other sources of systemic risk… but what is the key to recognizing a source of risk as being diversifiable or not. I don’t have a problem with systemic risks getting a premium. The challenge I have is believing that the portion of volatility that is not explained by systemic risk factors represents stuff that we can’t use smarts and information to understand. For example, maybe I have a theory that if the senior management of a company is old, then they are more likely to die and there will be a sudden search for new management, which will leave the company keel-less for a while, and reduce expected return. Therefore I put in a factor of “average age of senior management” in my model to cover the risk of management dying. It may be uncorrelated to the market beta, or it might even be correlated but the regression will correct for that. Let’s assume for a moment that the theory is true, that older management increases the likelihood of chaos in management and lower returns while things are being worked out. Is this a diversifiable risk, and if so, why? If not, why would it be a risk without expected return. Anyway, I keep going over this in my head, and maybe someone else can help think it through.
the problem with PM theory is that its an axiomatic system that is based on assumptions which rarely exist in the real world, same with economics. The reality is arbitrage situations do exist, markets are not entirely efficient and they sure as hell are not rational. Anyway I beleive the scenario you present is most certainly a diversifiable risk, its just that other market participants would not make the same insights therefore the security would be mispriced. I think that there are assets out there that have risk return ratios that exceed what you would find on the SML otherwise all PM’s would have the same return disregarding fees and asset allocation. thats all i can say right now, i had a few pints and its late so my brain isnt working too well.
Almost by definition every factor in an APT model is a diversifiable risk. All you have done is identify the well-known “codger factor”. There are lots of academic papers on it, but most of them have never been published as the authors often die before they are completed.