Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: Re: seasonality

From   "Nick Cox" <>
To   <>
Subject   RE: st: Re: seasonality
Date   Sat, 2 Apr 2005 18:03:38 +0100

There are lots of questions here. I'll 
peel off one. With this approach you shouldn't 
want or need to apply any special test. The regression 
output surely provides enough material for inference. 


kelly johnson
> thank you for your reply. can you please elaborate a bit more on the 
> additive method to use for actual retruns. i've generated the 
> required dummy 
> variables. and i've regressed my return series on the 11 
> dummy variables. 
> how do i proceed from here? where do i add teh original mean 
> (because as it 
> is i'm using the original series in the regression). any help 
> would be 
> appreciated. also, is there any test for significance that 
> can be done to 
> test the seasonality? a chi square test of some sort?
Kit Baum 

> >The textbook treatment of seasonality in economic and 
> financial time series 
> >involves creating a set of seasonal dummies (you need 3 for 
> Q, 11 for M) 
> >and then regressing your series on that set of dummies. Add 
> the original 
> >mean back into the series and you have a deseasonalized 
> series of stock 
> >returns.
> >
> >One can deal with additive seasonality (which would be more 
> appropriate for 
> >your returns series) or multiplicative seasonality (which 
> would be more 
> >appropriate for levels series, such as stock prices). The latter is 
> >achieved by regressing log(Y) on the dummies, exponentiating 
> the residuals, 
> >and adding back the mean of the original series.

> >>thank you for your reply. i express my seasonality question 
> a bit clearer.
> >>suppose we are looking at five years of stock retruns data. 
> how can we 
> >>test
> >>for teh presence of seasonality in specific months?
> >>if i were to do the typical adjustment, i'd first 
> calculated the average
> >>over teh entire time period, and divide my returns by that 
> average. call 
> >>the
> >>new variable rtn_adj. then i would have to average, by 
> month, to get the
> >>seasonl adjsutement indices for each month (which should 
> sum up to 12). i
> >>can then retruns and multiply teh original data by each of 
> the appropriate
> >>seasonal adjsutemtn factor. My question: is there a simpler 
> was to do 
> >>this?
> >>also, how can i do f-test to test teh presence of 
> seasonality? to give 
> >>youa
> >>better idea, i'm attaching a sample of the data to give you 
> an idea what 
> >>i'm
> >>looking at.

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index