Dear all,
I'm currently writing my pre-master thesis. For my thesis, I have panel data and ran a Negative binomial regression analysis with fixed effects.
I controlled for, fixed model or random-effects model by using the Hausman test which reported p<0.01 therefore, I used the fixed-effects model.
I also ran a negative binomial regression analysis with a random-effects model on my data to account for time-invariant industry effects. But this reported different results which do not make my results robust.
My question is How do I need to interpret these results? Does this mean that my dependent variable (environmental innovation) changes over time or my independent variable (CEO age)? If someone could elaborate on my results as I am not getting much wiser from the internet, I'm unsure if I understand correctly and how to proceed to know my results aren't robust.
See the boldly marked results for the changing results I mean:
Table 2
Results of negative binomial regression with fixed effects for the relationship between CEO age, R&D spending, and environmental innovation.
----------------------------------------------------------------------------------------------------
Dependent variable R&D Spending Environmental Innovation
---------------------------------------------------- -----------------------------------------
Model Model 2 Model 3 Model 4 Model 5
R&D intensity 2.073*** 2.055*** 10.90* 10.86* 6.172
(0.000) (0.000) (0.033) (0.034) (0.275)
Firm size 0.368*** 0.361*** 0.157 0.170 0.0329
(0.000) (0.000) (0.120) (0.105) (0.794)
Return on assets -0.327*** -0.329*** -1.037 -1.036 -0.572
(0.000) (0.000) (0.336) (0.336) (0.608)
CEO tenure -0.0000298*** -0.0000291*** -0.0000887* -0.0000880* -0.0000646
(0.000) (0.000) (0.035) (0.037) (0.146)
CEO salary 0.208*** 0.244*** -0.0936 -0.129 -0.0660
(0.000) (0.000) (0.770) (0.696) (0.842)
CEO bonus -0.0210** -0.0199** -0.0694 -0.0699 -0.0606
(0.001) (0.002) (0.446) (0.444) (0.496)
CEO age -0.00333** 0.00629 0.00489
(0.007) (0.655) (0.723)
R&D spending 0.000156*
(0.037)
Constant -2.120*** -2.113*** -0.335 -0.586 0.221
(0.000) (0.000) (0.854) (0.757) (0.910)
----------------------------------------------------------------------------------------------------
Observations 4805 4805 863 863 863
Log lik. -18517.5 -18513.9 -651.8 -651.7 -649.7
chi2 2449.5 2477.9 11.08 11.30 16.22
----------------------------------------------------------------------------------------------------
p-values in parentheses
* p<0.05, ** p<0.01, *** p<0.001
Results of negative binomial regression with random effects:
--------------------------------------------------------------------
(1) (2) (3)
number of ~ number of ~ number of ~
--------------------------------------------------------------------
number of green pa~
L.rdintensity -4.700 -4.272 -5.647*
(0.075) (0.103) (0.043)
firmsize 0.357*** 0.392*** 0.310***
(0.000) (0.000) (0.000)
roa -0.490 -0.506 -0.337
(0.484) (0.465) (0.650)
ceo_tenure -0.000106** -0.000111*** -0.0000956**
(0.002) (0.001) (0.005)
ceo_salary 0.174 0.101 0.143
(0.417) (0.607) (0.492)
L.ceo_bonus -0.0503 -0.0563 -0.0470
(0.528) (0.480) (0.542)
Executive's Age 0.0261* 0.0244*
(0.023) (0.030)
L.Research and Dev~e 0.000135*
(0.031)
Constant -4.229*** -5.534*** -5.125***
(0.001) (0.000) (0.000)
--------------------------------------------------------------------
/
ln_r 0.576*** 0.582*** 0.600***
(0.000) (0.000) (0.000)
ln_s -1.638*** -1.583*** -1.596***
(0.000) (0.000) (0.000)
--------------------------------------------------------------------
Observations 4925 4925 4925
Log lik. -1105.3 -1102.6 -1100.5
chi2 68.68 74.66 79.82
--------------------------------------------------------------------
p-values in parentheses
* p<0.05, ** p<0.01, *** p<0.001
Many thanks in advance!
Related Posts with Struggle with interpretation of time-invariant industry effects
Find and Replace all text within brackets?Hi, I am working with a panel dataset whose variable names are very different for every wave of the …
Problems with Outreg2 for my dissertationHello, I am having some problems running my code in stata. The code is as follows: regress dependen…
help with egen commandI am trying to look at labor division by gender in a household. my data set has collected the member…
Issue with binary logistic regression results when a predictor has only one of the binary responses.I'm using binary logistic regression, and I have encountered a problem with a predictor which has al…
Lead variable in unbalanced panel dataHi All, I have annual data on the sales revenue of firms. I have the year variable as follows: year …
Subscribe to:
Post Comments (Atom)
0 Response to Struggle with interpretation of time-invariant industry effects
Post a Comment