Beta Break and Parameter Stability

Jfuster

New Member
Hello, I am currently doing research on beta break of the basic CAPM model for an Argentine ETF.

The intuition is that after the election of a liberal president in 2015, the equity market in Argentina changed, and so the beta to the world (I use MSCI World as market index)

What triggered this research is the basic chart of the two stocks
Screen Shot 2018-04-18 at 11.49.38.png and Screen Shot 2018-04-18 at 11.49.21.png
(Argentina) (world)

Screen Shot 2018-04-18 at 11.49.38.png


If we see that roughly in January of 2016, the Argentine ETF starts to grow and mantain a relatively growing pattern, which to my eye differs with the previous pattern. Therefore, I thought that an assessment of a market change could be done. I figured that a beta break analysis would be the best option to assess this change, but I am not sure if its optimal. (any recommendations?)

To do so, I get the excess weekly returns, both for the ARG etf and for the WORLD etf, and create a dummy interaction variable for a break to test the change in slope and intercept

My model looks as follows:

Arg excess return=alpha+alpha(Break) + beta(Rworld-rf)+beta(Rworld-rf)*(Break)


What occurs, however, is that the beta actually diminishes after that event, contrary to my intuition. And I do not now whether my logic is wrong, or my modelling is wrong. Do you know why this might be?

Also, I would be interested in formally assessing a massive increase in trading volume, but I do not know which test to use (basically, trading volume seems to be significantly less before the "break" or "event" than after.

Screen Shot 2018-04-18 at 12.07.08.png


I think from that picture we can see a clear increase in trading volume, and I do not know how to report on this formally with a statistical test or regression.

Basically, my issue is: Intuitively, I see the argentine ETF (reflecting the Argentine market) to have changed. However, I don't know if my methods for assessing this change are optimal. I would highly appreciate any recommendations on how to proceed or improve my models and research.

Best,

Juan
 
Top