STOCK MARKET: problem with time-series data, dealing with business calendar dates

#1
Hey guys,

I'm currently investigating the relationship between daily stock returns and daily trading volume. More specifically, I am examining the period 01/01/90 - 31/12/2011 for FTSE100.

As it is well known, stock market is closed on Saturday and Sunday. Therefore, I should count for the so-called "business calendar" dates.

For example, in data editor it goes like this:
04jan1990
05jan1990 --> Friday
08jan1990 --> Monday
... It basically "jumps" from Friday to Monday.

Therefore, I cannot perform specific tests such as the Augmented Dickey-Fuller using 4 lags (eg. using the command: dfuller returns, trend lags(4)) as I get the error of "no observations".


I am aware of STATA providing specific commands to account for this, but it's only in Version 12 and I'm currently using Version 11.

Is there any alternative way to fix this problem?

Thank you.
 

bukharin

RoboStataRaptor
#2
The general principle is to create your own time scale, eg:
date mytime
04jan1990 1
05jan1990 2
08jan1990 3

Then you just use "mytime" as your time scale when you -tsset-