Greetings, everyone.
I am new to Statalist and this is my first post. Please forgive me for any mistakes in my English, as it is not my native language. I am new at GMM and I have encountered some difficulties in applying it to my data. I would appreciate your guidance on the suitability of using a time dummy and time effects in the system GMM on my data. My panel data sample consists of N= 1408; T=11 observations. The independent variables in the dynamic panel data model are some financial variables in time t; and the interaction of a family firm dummy with the lagged dependent variable. Additionally, I include an interaction between the lagged dependent variable with the family firm dummy and another dummy (dmy_1) of interest in my research, simultaneously.
My original model is:
#
According to literature, all independent financial variables are treated as endogenous so I lagged them accordingly.
I have the following questions:
1) The relevant literature suggests that for my variable of interest, it is important to account for firm-constant time effects. Is it appropriate to use a crisis dummy (it refers to two consecutive years corresponding to a financial crisis in my country) and, at the same time, include the years that do not refer to the crisis period (to account for firm-constant time effects)? Because if I include it together with "i.t" I have a drop problem due to collinearity since i.t includes the crisis period.
So, in this case, my new model would be:
#
?
2) I noticed that when including "i.t" in the first part of my model, I must also include "i.t" inside iv (i.t), so I assumed that I must also include the crisis dummy (like I did in my original model) if I use it. Is this correct?
3) Some authors use industry-fixed effects as an instrumental variable in the related literature. In that case, would it be suitable to use this instrumental variable together in iv()?
4) Could you explain what is the difference between using "gmm(Clw, lag (2 2)) gmm(Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1 , lag (2 2)" and "gmm (Clw Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1 , lag (2 2) eq(d))"? Can I infer that in the first case I am using the instrument "Clw, lag(2 2)" for the equations in levels, and all the right-hand side variables in the models lagged as instruments for the equations in differences?
My results are not supported by literature for almost all independent variables, that were not significant.
I apologize for the questions if they are unclear or too basic.
My results for the original model are:
#
#
I am new to Statalist and this is my first post. Please forgive me for any mistakes in my English, as it is not my native language. I am new at GMM and I have encountered some difficulties in applying it to my data. I would appreciate your guidance on the suitability of using a time dummy and time effects in the system GMM on my data. My panel data sample consists of N= 1408; T=11 observations. The independent variables in the dynamic panel data model are some financial variables in time t; and the interaction of a family firm dummy with the lagged dependent variable. Additionally, I include an interaction between the lagged dependent variable with the family firm dummy and another dummy (dmy_1) of interest in my research, simultaneously.
My original model is:
#
Code:
xtabond2 Clw l.Clw Famp Crisis_dmy dmy_1 c.l.Clw#c.Famp#c.dmy_1 Nww Cw Lvw Mk1w Cfw Szw , gmm(Clw, lag (2 2)) gmm(Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1 , lag (2 2) eq(d)) iv (i.StEconumber Crisis_dmy, eq(level)) two robust ortho small
I have the following questions:
1) The relevant literature suggests that for my variable of interest, it is important to account for firm-constant time effects. Is it appropriate to use a crisis dummy (it refers to two consecutive years corresponding to a financial crisis in my country) and, at the same time, include the years that do not refer to the crisis period (to account for firm-constant time effects)? Because if I include it together with "i.t" I have a drop problem due to collinearity since i.t includes the crisis period.
So, in this case, my new model would be:
#
Code:
xtabond2 Clw l.Clw Famp Crisis_dmy dmy_1 c.l.Clw#c.Famp#c.dmy_1 Nww Cw Lvw Mk1w Cfw Szw t1 t2 t3 t4 t7 t8 t9 t10 t11 , gmm(l.Clw, lag (2 2)) gmm(Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1 , lag (2 2) eq(d)) iv (i.StEconumber Crisis_dmy t1 t2 t3 t4 t7 t8 t9 t10 t11, eq(level)) two robust ortho small
2) I noticed that when including "i.t" in the first part of my model, I must also include "i.t" inside iv (i.t), so I assumed that I must also include the crisis dummy (like I did in my original model) if I use it. Is this correct?
3) Some authors use industry-fixed effects as an instrumental variable in the related literature. In that case, would it be suitable to use this instrumental variable together in iv()?
4) Could you explain what is the difference between using "gmm(Clw, lag (2 2)) gmm(Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1 , lag (2 2)" and "gmm (Clw Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1 , lag (2 2) eq(d))"? Can I infer that in the first case I am using the instrument "Clw, lag(2 2)" for the equations in levels, and all the right-hand side variables in the models lagged as instruments for the equations in differences?
My results are not supported by literature for almost all independent variables, that were not significant.
I apologize for the questions if they are unclear or too basic.
My results for the original model are:
#
Code:
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm. Warning: Two-step estimated covariance matrix of moments is singular. Using a generalized inverse to calculate optimal weighting matrix for two-step estimation. Difference-in-Sargan/Hansen statistics may be negative. Dynamic panel-data estimation, two-step system GMM ------------------------------------------------------------------------------ Group variable: i Number of obs = 1408 Time variable : t Number of groups = 195 Number of instruments = 109 Obs per group: min = 1 F(11, 194) = 95.68 avg = 7.22 Prob > F = 0.000 max = 10
#
Code:
Instruments for orthogonal deviations equation
GMM-type (missing=0, separate instruments for each period unless collapsed)
L2.(Famp Nww Cw Lvw Mk1w Cfw Szw dmy_1)
L2.Clw
Instruments for levels equation
Standard
1b.StEconumber 2.StEconumber 3.StEconumber 4.StEconumber 5.StEconumber
6.StEconumber 7.StEconumber 8.StEconumber 9.StEconumber 10.StEconumber
11.StEconumber 12.StEconumber 13.StEconumber 14.StEconumber 15.StEconumber
16.StEconumber 17.StEconumber 18.StEconumber 19.StEconumber 20.StEconumber
Crisis_dmy
_cons
GMM-type (missing=0, separate instruments for each period unless collapsed)
DL.Clw
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -3.07 Pr > z = 0.002
Arellano-Bond test for AR(2) in first differences: z = -1.00 Pr > z = 0.319
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(97) = 319.07 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(97) = 102.99 Prob > chi2 = 0.319
(Robust, but weakened by many instruments.)
Difference-in-Hansen tests of exogeneity of instrument subsets:
GMM instruments for levels
Hansen test excluding group: chi2(88) = 82.70 Prob > chi2 = 0.639
Difference (null H = exogenous): chi2(9) = 20.29 Prob > chi2 = 0.016
gmm(Clw, lag(2 2))
Hansen test excluding group: chi2(79) = 77.16 Prob > chi2 = 0.538
Difference (null H = exogenous): chi2(18) = 25.83 Prob > chi2 = 0.104
gmm(Fap Nww Cw Lvw Mk1w Cfw Szw dmy_1, eq(diff) lag(2 2))
Hansen test excluding group: chi2(25) = 24.94 Prob > chi2 = 0.466
Difference (null H = exogenous): chi2(72) = 78.05 Prob > chi2 = 0.292
iv(1b.StEconumber 2.StEconumber 3.StEconumber 4.StEconumber 5.StEconumber 6.StEconumber 7.StEco
> number 8.StEconumber 9.StEconumber 10.StEconumber 11.StEconumber 12.StEconumber 13.StEconumber
> 14.StEconumber 15.StEconumber 16.StEconumber 17.StEconumber 18.StEconumber 19.StEconumber 20.St
> Econumber Crisis_dmy, eq(level))
Hansen test excluding group: chi2(79) = 87.00 Prob > chi2 = 0.252
Difference (null H = exogenous): chi2(18) = 15.99 Prob > chi2 = 0.593
Comment