ppmlhdfe no standard errors reported

[Ann Suo]

Hello,

I am Ann Suo, when I use ppmlhdfe to deal with patent data with a lot of interactions, no standard errors reported. My data structure is as below:

id_firm year # of patents (per year) 5 wage variables(change in firm, year) country(change in firm) max_weight (change in firm) *numerical (73232+ lines)

There are 4899 firms,41 different countries and 15 years, totally 73232 observations. I want to apply country-year fixed effect weighted by max_weight, so I write codes below：

foreach xxx in AU AT BE BG BR CA CN CY CZ DE DK ES EE FI FR GB GR HU ID IN IE IT JP KR LT LU LV MX MT NL PL PT RO RU SK SI SE TR TW US {
forvalues yyy = 1995(1)2009 {
gen CY_`xxx'_`yyy' = 0
replace CY_`xxx'_`yyy' = max_weight if country == "`xxx'" & year == `yyy'
}
}
sort id_firm year

ppmlhdfe # of patents wage1 wage2 wage3 wage4 wage5 CY_* , absorb(id_firm) vce(cluster id_firm)

And after several minutes, dropping a lot of interactions, I got coefficients but no one standard errors.

How should I deal with this?

Thanks!

[Sergio Correia]

Two things:

1. The entire point of absorb() is for you to put the fixed effects there, but it seems you are creating hundreds of variables with country-year dummies where you could just have "a(id_firm country#year)".
2. I don't understand what max_weight is, or how can you "weight" fixed effects. I haven't heard of anything like that before, so giving more detail about the why would be useful.