Hi everyone!
I have panel data of the form:
I am trying to pinpoint the effects of drought on agricultural production and have a model of the form:
I am fine with the country FEs and the year FEs which I use to account for omitted variable bias. As such, I eliminate bias from time-varying unobservables that are constant across countries as well as eliminate bias due to factors that differ across countries but are constant over time (?).
Now, my issues:
1) I have been advised to include continent*year FEs instead of simple year FEs and I am struggling with the intuition behind this. I understand that I am in effect disaggregating the effects of year into separate year dummies for each continent and thus assuming that the time-varying unobvservables vary differently according to different continents... But I still am shaky about my understanding, if someone could clarify this and maybe have a pertinent example of the effects that are controlled for.
I've included these continent*year effects using reghdfe ln_agprod drought, absorb(country continent#year). Is that correct?
2) I would like to take into account time trends to see if I get different results. I am not interested in the coefficients for the trends but this would only be used to see the impact on the coefficient of drought. First of all, I assume I cannot include time fixed effects if I am including a time trend in my model (?). What is the difference between having time fixed effects and a time trend? I assume the trend incorporates the year effects?
I was advised to include country-specific time trends. So in my mind, I would include these country-specific time trends as well as country FEs but not year FEs.
Any idea of how I can do so in Stata? Would reghdfe ln_agprod drought, absorb(i.country##c.year) do the trick? or reg agprod spei i.country##c.year ?
Also has a clear way to explain the intuition between a model with 1) country fixed effects and year fixed effects only, 2) country fixed effects and continent*year fixed effects, and 3) country fixed effects and country-specific time trends? I feel a bit shaky about my understanding and would like to feel confident about what variation I am taking account for in each case.
Many many thanks in advance!
I have panel data of the form:
| Country | Continent | Year | agprod | drought |
| ABW | LAC | 1972 | 13413513 | 0 |
| ABW | LAC | 1973 | 15213523 | 1 |
| ABW | LAC | 1974 | 13513513 | 1 |
| ADO | WEOFF | 1972 | 353413 | 0 |
| ADO | WEOFF | 1973 | 234233 | 0 |
I am fine with the country FEs and the year FEs which I use to account for omitted variable bias. As such, I eliminate bias from time-varying unobservables that are constant across countries as well as eliminate bias due to factors that differ across countries but are constant over time (?).
Now, my issues:
1) I have been advised to include continent*year FEs instead of simple year FEs and I am struggling with the intuition behind this. I understand that I am in effect disaggregating the effects of year into separate year dummies for each continent and thus assuming that the time-varying unobvservables vary differently according to different continents... But I still am shaky about my understanding, if someone could clarify this and maybe have a pertinent example of the effects that are controlled for.
I've included these continent*year effects using reghdfe ln_agprod drought, absorb(country continent#year). Is that correct?
2) I would like to take into account time trends to see if I get different results. I am not interested in the coefficients for the trends but this would only be used to see the impact on the coefficient of drought. First of all, I assume I cannot include time fixed effects if I am including a time trend in my model (?). What is the difference between having time fixed effects and a time trend? I assume the trend incorporates the year effects?
I was advised to include country-specific time trends. So in my mind, I would include these country-specific time trends as well as country FEs but not year FEs.
Any idea of how I can do so in Stata? Would reghdfe ln_agprod drought, absorb(i.country##c.year) do the trick? or reg agprod spei i.country##c.year ?
Also has a clear way to explain the intuition between a model with 1) country fixed effects and year fixed effects only, 2) country fixed effects and continent*year fixed effects, and 3) country fixed effects and country-specific time trends? I feel a bit shaky about my understanding and would like to feel confident about what variation I am taking account for in each case.
Many many thanks in advance!
Comment