Beta drift

S2000magician wrote:
In which topic area / study session do you find this?
Click to expand...
It’s in the first reading on equities.
But to answer the question, when you are to estimate beta, analyst tend to add a beta drift/adjustment called the blume adjustment. This takes what your “long term beta” would be, which is one, and your estimated beta and creates a some what of weighted average putting 2/3 of weight towards your estimated beta and 1/3 towards the long term beta.
 
It assumes that a company’s long term beta is 1… So the adjusted beta is calculated as Adjusted beta = 0.33 + 0.67 (beta) which gives a less biased estimate of long term beta. Beta is considered mean reverting to 1 so with a beta above one that equation gives a result closer to one, and with a beta below one is also gives a beta closer to one.
Adjusted beta = 0.33 + 0.67 (beta) is a standard treatment, although it isn’t set in stone. It depends how quickly you think beta will mean revert to 1. For example if you think it will mean revert more quickly, you could use Adjusted beta = 0.5 + 0.5 (beta)
It’s important to note that this concept isn’t beta “drift” it is beta “mean reversion”. Two very different things!!!
 
this somewhat concides with the quant section with random walk. So let’s say (and this is question rather than a statement) but if you don’t have a mean reversion, that means you have a unit root, and that means it is not stationary. Yeah?
 
badger2014 wrote:
When do you use beta drift and why is it useful?
Click to expand...
You have to use it in case you are forecasting a beta you have generated from a regression analysis (with historical inputs).
 
You must log in or register to reply here.
Back
Top