Hey everyone,
I am estimating the following model consisting of 29 OECD-countries over 23 years
lnC = lnP+lnY+lnU+lim+com
The overall objective is to determine if an advertising bans reduces tobacco cosumption.
lnC = tobacco consumption
lnP = price
lnY =income
lnU = unemployment
and an advertising restriction score weak, limited and comprehensive, whereby only lim (limited) and com (comprehensive) are included due to multicollinearity!
I am using the augmented mean group estimator designed for macro panel data by Markus Eberhardt that allows for slope heterogeneity, non-stationarity and cross-sectional dependence.
. xtmg logC logP logY logU lim com, augment trend
Augmented Mean Group estimator (Bond & Eberhardt, 2009; Eberhardt & Teal, 2010)
Common dynamic process included as additional regressor
All coefficients represent averages across groups (group variable: Country)
Coefficient averages computed as unweighted means
Mean Group type estimation Number of obs = 586
Group variable: Country Number of groups = 28
Obs per group: min = 16
avg = 20.9
max = 23
Wald chi2(5) = 12.00
Prob > chi2 = 0.0348
logC Coef. Std. Err. z P>z [95% Conf. Interval]
logP -.1084904 .0550959 -1.97 0.049 -.2164763 -.0005044
logY .355781 .3058294 1.16 0.245 -.2436335 .9551956
logU -.0080689 .0470321 -0.17 0.864 -.1002501 .0841122
lim -.0199633 .0133105 -1.50 0.134 -.0460515 .0061249
com .0111684 .0168496 0.66 0.507 -.0218562 .0441929
c_d_p .644051 .2930564 2.20 0.028 .069671 1.218431
trend -.0120807 .0127358 -0.95 0.343 -.0370423 .012881
_cons 4.174708 3.136262 1.33 0.183 -1.972253 10.32167
Root Mean Squared Error (sigma): 0.0498
Variable c_d_p refers to the common dynamic process.
Variable trend refers to the group-specific linear trend terms.
Share of group-specific trends significant at 5% level: 0.286 (= 8 trends)
I do have difficulties understanding, what exactly does the "c_d_p" and the "trend" variable tell me in this context. What is the difference between them? CDP is the common dynamic process, that accounts for common time-varying effect in all OECD countries, right?
Is it recommendable to include "trend" or "impose" in the command?
Thanks a lot in advance!
Best regards,
Louisa
I am estimating the following model consisting of 29 OECD-countries over 23 years
lnC = lnP+lnY+lnU+lim+com
The overall objective is to determine if an advertising bans reduces tobacco cosumption.
lnC = tobacco consumption
lnP = price
lnY =income
lnU = unemployment
and an advertising restriction score weak, limited and comprehensive, whereby only lim (limited) and com (comprehensive) are included due to multicollinearity!
I am using the augmented mean group estimator designed for macro panel data by Markus Eberhardt that allows for slope heterogeneity, non-stationarity and cross-sectional dependence.
. xtmg logC logP logY logU lim com, augment trend
Augmented Mean Group estimator (Bond & Eberhardt, 2009; Eberhardt & Teal, 2010)
Common dynamic process included as additional regressor
All coefficients represent averages across groups (group variable: Country)
Coefficient averages computed as unweighted means
Mean Group type estimation Number of obs = 586
Group variable: Country Number of groups = 28
Obs per group: min = 16
avg = 20.9
max = 23
Wald chi2(5) = 12.00
Prob > chi2 = 0.0348
logC Coef. Std. Err. z P>z [95% Conf. Interval]
logP -.1084904 .0550959 -1.97 0.049 -.2164763 -.0005044
logY .355781 .3058294 1.16 0.245 -.2436335 .9551956
logU -.0080689 .0470321 -0.17 0.864 -.1002501 .0841122
lim -.0199633 .0133105 -1.50 0.134 -.0460515 .0061249
com .0111684 .0168496 0.66 0.507 -.0218562 .0441929
c_d_p .644051 .2930564 2.20 0.028 .069671 1.218431
trend -.0120807 .0127358 -0.95 0.343 -.0370423 .012881
_cons 4.174708 3.136262 1.33 0.183 -1.972253 10.32167
Root Mean Squared Error (sigma): 0.0498
Variable c_d_p refers to the common dynamic process.
Variable trend refers to the group-specific linear trend terms.
Share of group-specific trends significant at 5% level: 0.286 (= 8 trends)
I do have difficulties understanding, what exactly does the "c_d_p" and the "trend" variable tell me in this context. What is the difference between them? CDP is the common dynamic process, that accounts for common time-varying effect in all OECD countries, right?
Is it recommendable to include "trend" or "impose" in the command?
Thanks a lot in advance!
Best regards,
Louisa
