st: RE: lag

[Nick Cox <n.j.cox@durham.ac.uk>]

If -sale[_n-4]- is not defined because _n-4 <= 1 then Stata evaluates -sale[_n-4]- as missing, meaning . or system missing. Only values of .a ... .z can be greater than that. 

This may be clearer if your code is re-written as one statement: 

by cusip: generate sale_decrease5 = sale <= sale[_n-4]

What is being evaluated is the expression 

sale <= sale[_n-4] 

and this will be returned as 1 or 0, depending on whether it is true. Missing values of either argument don't affect that. 

This is on all fours with (say) 

. di (42 <= .)
1

If you want your new variable to be . if either argument is missing you can do this 

by cusip: generate sale_decrease5 = sale <= sale[_n-4] if !missing(sale, sale[_n-4])  

and then your indicator will have possible values 0, 1 and missing. 

You could also do this: 

by cusip: generate sale_decrease5 = sale <= sale[_n-4] if _n > 4 

See also 

FAQ     . . . . . . . . . . . . . . . . . . . . . . .  True and false in Stata
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
        8/05    What is true and false in Stata?
                http://www.stata.com/support/faqs/data/trueorfalse.html

Nick 
n.j.cox@durham.ac.uk 

Fabian Schönenberger

With the following command I want to generate a dummy variable; 1 if current sales are lower than 4 years ago, 0 if sales are larger for each company:

by cusip: generate sale_decrease5=0 if sale>sale[_n-4]
by cusip: replace sale_decrease5=1 if sale<=sale[_n-4]

I expected for each cusip 4 missings at the beginning, since there is no _n-4. However, the command gives 1 for each cusip in the first four years. 

How do I have to adapt the formula to get dummy variables after the first 4 years passed?

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/