Country Fixed Effects

[Edward Wu]

Hmmm...an easy question

How could I generate country fixed effects, based on a string variable (loc) that contains country names?

Tried run

xtreg .... i.loc ...

didn't work. Destring...generate () force also didn't work.

Thank you!

[Carlo Lazzaro]

Edward:
-i.loc- with a string variable won't work.
Then you should convert it in numeric format, as in the following toy-example:

. set obs 3
number of observations (_N) was 0, now 3

. g loc="Rome" in 1
(2 missing values generated)

. replace loc="Paris" in 2
variable loc was str4 now str5
(1 real change made)

. replace loc="London" in 3
variable loc was str5 now str6
(1 real change made)

. g count_city=_n

. label define count_city 1 "Rome" 2 "Paris" 3 "London"

. label val count_city count_city

. list loc count_city

     +-------------------+
     |    loc   count_~y |
     |-------------------|
  1. |   Rome       Rome |
  2. |  Paris      Paris |
  3. | London     London |
     +-------------------+
  

In you future post, please use -dataex- to provide an excerpt/example of your data, so that interested listers can give you more helpful replies. Thanks.

[Phil Bromiley]

Alternatively

egen countrynum=group(country)

will assign a number for each country. Then i.countrynum works for the dummies. Alternatively, you can use it in the xtset.

[Edward Wu]

Thanks both!! It worked! Have a wonderful weekend!