I seek to understand the difference between unit / entity / individual fixed effects and time fixed effects. Both are oneway.

Many questions surround them in contrast or addition to twoway fixed effects, which limit interpretation and may have other flaws (suggested reading http://web.mit.edu/insong/www/pdf/FEmatch-twoway.pdf).

So, I like to stick to the oneway effects. Here is a code example:

soep <- read_dta("https://github.com/MarcoKuehne/marcokuehne.github.io/blob/main/data/SOEP/practice_en/practice_dataset_eng.dta?raw=true")

FE_unit <- plm(erwerb ~ bildung + alter + anz_kind, 
    model = "within",
    effect = "individual",
    index=c("id", "syear"))

FE_time <- plm(erwerb ~ bildung + alter + anz_kind, 
               model = "within",
               effect = "time",
               index=c("id", "syear"))

stargazer(FE_unit, FE_time, type="text")

                               Dependent variable:                  
                        (1)                         (2)             
bildung              -0.402***                   -0.127***          
                      (0.020)                     (0.004)           
alter                  -0.001                     0.037***          
                      (0.004)                     (0.001)           
anz_kind              0.197***                   -0.049***          
                      (0.019)                     (0.011)           
Observations           21,851                      21,851           
R2                     0.033                       0.171            
Adjusted R2            -0.320                      0.171            
F Statistic  182.071*** (df = 3; 16001) 1,504.182*** (df = 3; 21843)
Note:                                    *p<0.1; **p<0.05; ***p<0.01

What does it mean for age to have a significant time fixed effects whereas it has no unit (individual) fixed effect on employment?

I have seen the underlying formulas but am interested in use cases and applications. One of the often cited sources is https://www.econometrics-with-r.org/10-4-regression-with-time-fixed-effects.html Controlling for variables that are constant across entities but vary over time can be done by including time fixed effects. Can you please give me an illustrative example? Can you name publications focusing on time fixed effects?

The best example I could find is this https://byelenin.github.io/MicroEconometrics/Slides/GradMetrics_2020_Lec7A.pdf Like safety improvements in new cars as an omitted variable that changes over time but has the same value for all states.




Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.