For a project we have to use performance regression to estimate whether a portfolio delivers a significant alpha. We should run the following time series regression for this portfolio:
rβL,t = αβL,t + γ1,βL,t(Mkt − Rf)t + γ2,βL,tSMBt + γ3,βL,tHMLt + eβL,t
I'm already stuck with doing the time series regression. I tried it like this (i'm really not good in statistics, and even worse at stata, which makes it hard to gauge if this even makes sense):
(PF = Portfolio, mkt_fr = Mkt-RF)
Code:
gen PF = (P) gen mkt_rf = MktRF - RF *regress return_lowP mkt_rf SMB HML if PF == 5, robust cap regress return_lowP mkt_rf SMB HML, robust gen nobs = e(N) gen b_MktRF = _b[mkt_rf] gen t_MktRF = _b[mkt_rf] / _se[mkt_rf] gen b_SMB = _b[SMB] gen t_SMB = _b[SMB] / _se[SMB] gen b_HML = _b[HML] gen t_HML = _b[HML] / _se[HML] gen b_cons = _b[_cons] gen t_cons = _b[_cons] / _se[_cons] gen r2 = e(r2) gen r2a = e(r2_a) runby myreg, by(PF)
--------------------------------------
Number of by-groups = 25
by-groups with errors = 25
by-groups with no data = 0
Observations processed = 27,900
Observations saved = 0
--------------------------------------
Could someone help me out on how to approach this task?
Many thanks in advance!
0 Response to time series regressions for a portfolio
Post a Comment