Hi!
I'm trying to analyze whether Government expenditures have the potential to reduce income inequality in a Dynamic Panel data of 30 countries (1995-2020).
I first computed fixed effects estimators, but I want to add system GMM as robustness check as I want to avoid endogeneity issues.
It's the first time I use -xtabond2- so ran multiple commands. The following one yields the "best" results for now:
With top 10% income share post transfers as my dependent variable, different government expenditure types and a set of covariates (GDP per capita, human capital, unemployment, taxes on products, income taxes, openness to trade and inflation).
I use the -orthogonal- option as my panel is unbalanced.
This is the result:
1) Obviously I have too many instrument although I used the -collapse- command. Is there any other way to reduce the number of instruments?
2) Is there also a way to improve the model as to increasing the significance of the coefficients? My fixed effect estimators yielded better results. I am also quite surprised that all government covariates (except for total government expenditures) have a positive relationship with income inequality; this in not the case in the literature nor in my FE analysis.
Thanks a lot for your help!
I'm trying to analyze whether Government expenditures have the potential to reduce income inequality in a Dynamic Panel data of 30 countries (1995-2020).
I first computed fixed effects estimators, but I want to add system GMM as robustness check as I want to avoid endogeneity issues.
It's the first time I use -xtabond2- so ran multiple commands. The following one yields the "best" results for now:
Code:
xtabond2 top10sharePost l.top10sharePost total_GE educ_GE health_GE defence_GE social_prot_GE logGDPpc c.logGDPpc#c.logGDPpc hc Unempl TaxP TaxIn Trade CPI i.Year, gmm(l.top10sharePost total_GE educ_GE health_GE defence_GE social_prot_GE, collapse) iv(logGDPpc c.logGDPpc#c.logGDPpc hc Unempl i.Year) robust orthogonal small
I use the -orthogonal- option as my panel is unbalanced.
This is the result:
Code:
xtabond2 top10sharePost l.top10sharePost total_GE educ_GE health_GE defence_GE social_prot_GE logGDPpc c.logGDPpc#c.logGDPp > c hc Unempl TaxP TaxIn Trade CPI i.Year, gmm(l.top10sharePost total_GE educ_GE health_GE defence_GE social_prot_GE, collaps > e) iv(logGDPpc c.logGDPpc#c.logGDPpc hc Unempl i.Year) robust orthogonal small Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm. 1995b.Year dropped due to collinearity 2017.Year dropped due to collinearity 2020.Year dropped due to collinearity Warning: Number of instruments may be large relative to number of observations. Warning: Two-step estimated covariance matrix of moments is singular. Using a generalized inverse to calculate robust weighting matrix for Hansen test. Difference-in-Sargan/Hansen statistics may be negative. Dynamic panel-data estimation, one-step system GMM ------------------------------------------------------------------------------ Group variable: code Number of obs = 688 Time variable : Year Number of groups = 30 Number of instruments = 177 Obs per group: min = 5 F(37, 29) = 713.94 avg = 22.93 Prob > F = 0.000 max = 24 --------------------------------------------------------------------------------------- | Robust top10sharePost | Coef. Std. Err. t P>|t| [95% Conf. Interval] ----------------------+---------------------------------------------------------------- top10sharePost | L1. | .6949555 .0602359 11.54 0.000 .5717593 .8181518 | total_GE | -.0011993 .0004547 -2.64 0.013 -.0021292 -.0002694 educ_GE | .0031353 .0035082 0.89 0.379 -.0040397 .0103103 health_GE | .0005659 .0023851 0.24 0.814 -.0043122 .005444 defence_GE | .0045551 .0044644 1.02 0.316 -.0045756 .0136858 social_prot_GE | .0000675 .0011989 0.06 0.956 -.0023846 .0025195 logGDPpc | .0407186 .0447143 0.91 0.370 -.0507324 .1321696 | c.logGDPpc#c.logGDPpc | -.0020422 .0022727 -0.90 0.376 -.0066904 .002606 | hc | -.0120358 .0056354 -2.14 0.041 -.0235615 -.0005101 Unempl | .0005721 .0004153 1.38 0.179 -.0002772 .0014214 TaxP | .000132 .0015373 0.09 0.932 -.0030121 .0032762 TaxIn | -.00012 .0012538 -0.10 0.924 -.0026842 .0024443 Trade | 8.41e-06 .0000392 0.21 0.832 -.0000718 .0000886 CPI | -.0000104 .0000632 -0.16 0.870 -.0001398 .0001189 | Year | 1996 | -.0099925 .0066345 -1.51 0.143 -.0235615 .0035766 1997 | -.0048797 .0061826 -0.79 0.436 -.0175245 .0077652 1998 | -.0066666 .0066803 -1.00 0.327 -.0203294 .0069963 1999 | -.010615 .0058941 -1.80 0.082 -.0226699 .0014398 2000 | -.0048585 .0053659 -0.91 0.373 -.015833 .006116 2001 | -.0097027 .0047246 -2.05 0.049 -.0193656 -.0000398 2002 | -.0075007 .005241 -1.43 0.163 -.0182197 .0032183 2003 | -.0089584 .0048729 -1.84 0.076 -.0189246 .0010079 2004 | -.0068912 .0036733 -1.88 0.071 -.014404 .0006216 2005 | -.0038052 .0045227 -0.84 0.407 -.0130552 .0054448 2006 | -.0036164 .0049765 -0.73 0.473 -.0137945 .0065617 2007 | -.0002656 .0052403 -0.05 0.960 -.0109832 .010452 2008 | -.0087284 .0034038 -2.56 0.016 -.0156899 -.0017668 2009 | -.0068449 .0042478 -1.61 0.118 -.0155325 .0018428 2010 | -.0025293 .0050697 -0.50 0.622 -.0128981 .0078395 2011 | -.0026926 .0037836 -0.71 0.482 -.0104309 .0050456 2012 | -.0016278 .0026139 -0.62 0.538 -.0069737 .0037181 2013 | -.0011289 .0048741 -0.23 0.818 -.0110975 .0088397 2014 | .0023757 .0026167 0.91 0.371 -.002976 .0077275 2015 | -.006012 .0043892 -1.37 0.181 -.0149889 .0029648 2016 | -.0043263 .0028378 -1.52 0.138 -.0101302 .0014776 2018 | -.0002148 .0030223 -0.07 0.944 -.006396 .0059665 2019 | -.0009046 .0033603 -0.27 0.790 -.0077773 .005968 | _cons | -.0523268 .2136437 -0.24 0.808 -.4892773 .3846237 --------------------------------------------------------------------------------------- Instruments for orthogonal deviations equation Standard FOD.(logGDPpc c.logGDPpc#c.logGDPpc hc Unempl 1995b.Year 1996.Year 1997.Year 1998.Year 1999.Year 2000.Year 2001.Year 2002.Year 2003.Year 2004.Year 2005.Year 2006.Year 2007.Year 2008.Year 2009.Year 2010.Year 2011.Year 2012.Year 2013.Year 2014.Year 2015.Year 2016.Year 2017.Year 2018.Year 2019.Year 2020.Year) GMM-type (missing=0, separate instruments for each period unless collapsed) L(1/25).(L.top10sharePost total_GE educ_GE health_GE defence_GE social_prot_GE) collapsed Instruments for levels equation Standard logGDPpc c.logGDPpc#c.logGDPpc hc Unempl 1995b.Year 1996.Year 1997.Year 1998.Year 1999.Year 2000.Year 2001.Year 2002.Year 2003.Year 2004.Year 2005.Year 2006.Year 2007.Year 2008.Year 2009.Year 2010.Year 2011.Year 2012.Year 2013.Year 2014.Year 2015.Year 2016.Year 2017.Year 2018.Year 2019.Year 2020.Year _cons GMM-type (missing=0, separate instruments for each period unless collapsed) D.(L.top10sharePost total_GE educ_GE health_GE defence_GE social_prot_GE) collapsed ------------------------------------------------------------------------------ Arellano-Bond test for AR(1) in first differences: z = -3.50 Pr > z = 0.000 Arellano-Bond test for AR(2) in first differences: z = 1.36 Pr > z = 0.173 ------------------------------------------------------------------------------ Sargan test of overid. restrictions: chi2(139) = 225.95 Prob > chi2 = 0.000 (Not robust, but not weakened by many instruments.) Hansen test of overid. restrictions: chi2(139) = 0.00 Prob > chi2 = 1.000 (Robust, but weakened by many instruments.)
2) Is there also a way to improve the model as to increasing the significance of the coefficients? My fixed effect estimators yielded better results. I am also quite surprised that all government covariates (except for total government expenditures) have a positive relationship with income inequality; this in not the case in the literature nor in my FE analysis.
Thanks a lot for your help!
Comment