Including fixed year dummies in xtabond command

[Drishta Tengur]

I plan to implement the difference GMM estimator for a panel of 27 countries for the 1995-2015 period. However, when I include -i.year- to capture time specific effects, I get the following results:

 xtabond PATR1 preelection right left plurality system coalition pit corporate inflation gdp un
> employment govtexp cashtransfers wage waged urbanisation dependency dependencyold dependencyyo
> ung popbelow14 pop1564 pop65andabove i.year
factor variables not allowed
r(101);

Please help me understand what is wrong in the command?

[Carlo Lazzaro]

Drishta:
try again replacing

i.year

with:

yr1995-yr2015

[Drishta Tengur]

Carlo:
This is what I get now:

xtabond PATR1 preelection right left plurality system coalition pit corporate inflation gdp un
> employment govtexp cashtransfers wage waged urbanisation dependency dependencyold dependencyyo
> ung popbelow14 pop1564 pop65andabove yr1995-yr2015
variable yr1995 not found
r(111);

[Jeff Wooldridge]

Drishta: You will have to define the year dummies explicitly. So, explicitly define dummy variables named yr1995, yr1996, up to yr2015.

[Jon Lee]

Could someone please advise me whether you should include year dummies in your xtabond model and, if so, why?