I am an absolute beginner trying to run regressions and correctly intepret these regressions in Stata. Currently, I am running a regression using Swedish data where I am trying to establish if there's a correlation between an individual's marginal tax rate and that individual's propensity to use sick days. The hypothesis is that individuals on average would become more inclined to use sick days as their marginal tax rate increases since they get to keep a smaller piece of their income pie.
The data which I am using is panel data from 1988-1991 for more than 500,000 individuals. For each individual there is data that indicates the individual's marginal tax rate and number of sick days for every year.
Currently, I am running the following regression:
reg sickdays_all mrtax i.year
Where "sickdays_all" is the number of sick days used by an individual in a given year and "mrtax" is the percentage of income an individual has to pay in taxes. In addition to this I am using fixed effects for years, a concept with which I am only, at best, vaguely familiar.
The output of this regression looks as follows:
Linear regression
sickdays_all | Coef. | St.Err | t-value | p-value | Sig. | |||
mrtax | -0.286 | 0.006 | -51.43 | 0.000 | *** | |||
1988b.year | 0.000 | . | . | . | ||||
1989.year | -0.899 | 0.152 | -5.92 | 0.000 | *** | |||
1990.year | -0.179 | 0.153 | -1.17 | 0.241 | ||||
1991.year | -5.207 | 0.168 | -30.95 | 0.000 | *** | |||
_cons | 31.873 | 0.296 | 107.61 | 0.000 | *** | |||
Mean dependent var | 17.430 | SD dependent var | 38.868 | |||||
R-squared | 0.005 | Number of obs | 529461.000 | |||||
F-test | 725.900 | Prob > F | 0.000 | |||||
Akaike crit. (AIC) | 5375494.390 | Bayesian crit. (BIC) | 5375550.288 | |||||
*** p<0.01, ** p<0.05, * p<0.1 |
With that in mind, my questions are as follows:
1. My interpretation of the coefficient for mrtax is that, on average, a one percent increase in the marginal tax rate results in a decrease of 0.286 sick days. Is this interpretation correct?
2. For each year used in the fixed effects model there is a coefficient. How do I interpret this coefficient? What is the intuition behind it and what does it mean in this particular context?
I am a first-time poster and as such I probably did not conform fully to the standards of this forum. I apologize for this. I will be most grateful for any response I can get.
Respectfully,
Burre
0 Response to Regarding fixed effects
Post a Comment