First, make a declaration of the panel data:
In front is the cross-section element, followed by the time marker:
tsset company year
tsset industry year
Generate a new variable: gennewvar=human*lnrd
Produces the hysteresis variable genfiscal(2)=l2fiscal
Differential variable genfiscal(d)=d. is generatedfiscal
1. Descriptive statistics.
xtdes: the overall description of the number of cross-sections and time span of the panel data.
xtsum: Basic statistics for individual variables are calculated within groups, between groups, and samples as a whole.
xttab displays the distribution of a variable in a list.
II. Main Commands and Methods.
The main command in stata for estimating the panel model: xtreg
xtreg depvar [varlist] [if exp] ,model_type [level(#
model type.
be between-effects estimator
fe fixed-effects estimator
re glsrandom-effects estimator
pa geepopulation-**eraged estimator
mle maximum-likelihood random-effectsestimator
Main Estimation Methods:
xtreg： fixed-, between- and random-effects, and population-**eraged linear models
xtregar：fixed- andrandom-effects linear models with an ar(1) disturbance
xtpcse ：ols orprais-winsten models with panel-corrected standard errors
xtrchh ：hildreth-houckrandom coefficients models
xtivreg ：instrumentalvariables and two-stage least squares for panel-data models
xtabond：arellano-bond linear, dynamic panel data estimator
xttobit ：random-effectstobit models
xtlogit ：fixed-effects,random-effects, population-**eraged logit models
xtprobit ：random-effects andpopulation-**eraged probit models
xtfrontier ：stochastic frontiermodels for panel-data
xtrc gdp invest culture edu sci health social admin,beta
3. Application of the xtreg command.
Declaring Panel Data Types:
1. Panel Declaration.
use fdi.dtar, clear
xtset id year
1.Fixed-effect model estimates:
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,fe
2.Random-effects model estimates:
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,re
3.Maximum likelihood estimation ml:
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,mle
Hausman's test is whether to choose a fixed-effect model or a random-effects model
Step 1: Estimate the fixed-effect model and store the results.
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,fe
est store fe
Step 2: Estimate the random-effects model and store the results.
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,re
est store re
Step 3: Perform a Hausman test.
hausman fe re
Heteroskedasticity and sequence-correlation tests for fixed-effect models:
xtserial xtreg lngdp lnfdi lnie lnex lnim lnci lngp
Heteroskedasticity test:
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,fe
xttest3 (modified wald statistic for groupwise heteroskedasticity in fixedeffect model)
Sequence correlation tests for random-effects models:
xtreg xtreg lngdp lnfdi lnie lnex lnim lnci lngp,re
xttest1
XTTet1 was used to test for random effects (uni-tailed and two-tailed), first-order sequence associations, and a combination of the two.
The results of the test showed that there was a random effect and sequence correlation, and the combined test for random effect and sequence correlation was also very significant.
Heteroskedasticity and sequence correlation can be corrected using the generalized linear model XTGLS
xtgls xtreg lngdp lnfdi lnie lnex lnim lnci lngp, panels (hetero), corrected for heteroskedasticity.
xtgls xtreg lngdp lnfdi lnie lnex lnim lnci lngp, panels(correlated), correcting for heteroskedasticity depending on the cross-section.
xtgls xtreg lngdp lnfdi lnie lnex lnim lnci lngp, panels(hetero) corr(ar1), corrected heteroskedasticity and first-order sequence correlation ar(1).