Page 1 of 1
OLS with series containing different number of obs
Posted: Tue Jun 25, 2013 3:07 pm
by Lassebn
In my data set I have a dataseries which consists of real stock prices from 1979Q1-2012Q2 and a series for the P/E_ratio for 1980Q1-2012Q2
I want to run the following OLS:
log(real_stock_price)-log(real_stock_price(-1)) c log(P/E_ratio)
Written like this the dependent variable is the logaritmic real return (I want to construct it like this - not make a new series for real return from 1980Q1-2012Q2)
and because i write log(real_stock_price(-1)) i thought Eviews would use the observation from 1979Q1 but it does not..
Is it impossible for Eviews to use the 1979Q4 observation like this when the series have different numbers of observations??
Re: OLS with series containing different number of obs
Posted: Tue Jun 25, 2013 4:06 pm
by startz
In my data set I have a dataseries which consists of real stock prices from 1979Q1-2012Q2 and a series for the P/E_ratio for 1980Q1-2012Q2
I want to run the following OLS:
log(real_stock_price)-log(real_stock_price(-1)) c log(P/E_ratio)
Written like this the dependent variable is the logaritmic real return (I want to construct it like this - not make a new series for real return from 1980Q1-2012Q2)
and because i write log(real_stock_price(-1)) i thought Eviews would use the observation from 1979Q1 but it does not..
Is it impossible for Eviews to use the 1979Q4 observation like this when the series have different numbers of observations??
EViews will use the 1979Q4 to calculate log(real_stock_price)-log(real_stock_price(-1)) for 1980q1. Is that what you want?
Re: OLS with series containing different number of obs
Posted: Wed Jun 26, 2013 12:23 am
by Lassebn
In my data set I have a dataseries which consists of real stock prices from 1979Q1-2012Q2 and a series for the P/E_ratio for 1980Q1-2012Q2
I want to run the following OLS:
log(real_stock_price)-log(real_stock_price(-1)) c log(P/E_ratio)
Written like this the dependent variable is the logaritmic real return (I want to construct it like this - not make a new series for real return from 1980Q1-2012Q2)
and because i write log(real_stock_price(-1)) i thought Eviews would use the observation from 1979Q1 but it does not..
Is it impossible for Eviews to use the 1979Q4 observation like this when the series have different numbers of observations??
EViews will use the 1979Q4 to calculate log(real_stock_price)-log(real_stock_price(-1)) for 1980q1. Is that what you want?
Yes that is what i want.. Then why does Eviews write:
Sample (adjusted): 1980Q1 2012Q2
in my estimation output?
Re: OLS with series containing different number of obs
Posted: Wed Jun 26, 2013 6:28 am
by EViews Gareth
Because those are the observations used in the estimation, even if values from other dates are used in the calculation of those observations.
Re: OLS with series containing different number of obs
Posted: Thu Jun 27, 2013 4:56 am
by Lassebn
Thank you both :)