I am working with severely unbalanced Panel Data of a nations Fisheries where I have individual data from all deliveries made by every single vessel. Thus far I have reshaped the data so that every observation amounts to a single delivery by a single vessel on a single day. The reason I think they are unbalanced is that every time period consists of different vessels as they're not landing fish every day.
In Stata I use xtset
to define the individual vessels as my panels and use days as my time variable as they in certain periods will deliver as often as every day.
Seeing as I want to control for the vessels characteristics to explain my dependent variables(weight landed and quality on weight landed) I am using time invariant variables such as vessel length and engine power, in addition to a lot of time-varying variables.
Based on what I read I thought this meant that I had to use a Random Effects model in order for the time-invariant variables within panels not to be removed, but when running a Fixed effects regression Stata keeps all the variables. This leads me to think I might have specified the model wrongfully?
Running a Hausman test on the two regressions gives a large but negative chi-square statistic.
Additionally I'd like to control for differences in seasons, which I have done by adding quarterly dummies, and also differences in years by adding yearly dummies for the eight year period. These come out as statistically significant and plausible based on what I know about the fisheries. However, I am wondering if adding yearly/quarterly dummies is wrong as the Xtset command have already defined time dummies for each day?
Any input would be greatly appreciated. I have also included my lines of code below. The reason for included the clustering is heteroskedasticity.
xtset fartyid landingsdato_ny
xtreg ln_r_prisperkg_Frst_102202 Dflere_mottak_tur i.landingsfylkekode i.kvartiler_ny markedsk_torsk gjenv_TAC_NØtorsk_år_prct lalder_fartøy i.fangstr r_minst_Frst_torsk gjenv_kvote_NØtorsk_fartøy_prct i.lengde_gruppering mobilitet, re vce(cluster fartyid)
xtreg ln_r_prisperkg_Frst_102202 Dflere_mottak_tur i.landingsfylkekode i.kvartiler_ny markedsk_torsk gjenv_TAC_NØtorsk_år_prct lalder_fartøy i.fangstr r_minst_Frst_torsk gjenv_kvote_NØtorsk_fartøy_prct i.lengde_gruppering mobilitet, fe vce(cluster fartyid)