I am looking to create a correlation matrix to describe correlations between my variables in a panel dataset, taking into account both time variance and cross-sectional variance. In checking that I am using the write command, I have run across multiple different ways of getting a correlation matrix from panel data. These differences seemed to depend on whether one was using a fixed effects or random effects regression, but I am running PMG estimation on the error correction model with the stata command xtdcce2 seen below (excluding for simplicity the measures of tax in the correlation matrix):
The code I am using to get the correlation matrix, and the output, looks thus:
I just wanted to check I am not making a glaring error here. Is "corr" the right command in this context? Thanks!
Code:
xtdcce2 d.growth d.l(growth) d.l(0/1)(gfcfgdp employmentgrowth yearsedugrowth l.lngdpph) if year < 2007, ///
lr(l.growth gfcfgdp employmentgrowth yearsedugrowth l.lngdpph) ///
p(l.growth gfcfgdp employmentgrowth yearsedugrowth l.lngdpph) nocross exponent
Code:
estpost corr deltagrowth tau_k tau_h tau_c topstat corptax, matrix
| e(b) e(rho) e(p) e(count)
-------------+--------------------------------------------
deltagrowth |
deltagrowth | 1 1 624
tau_k | -.0242413 -.0242413 .550123 610
tau_h | .0054273 .0054273 .8935842 610
tau_c | -.0135446 -.0135446 .7378667 613
topstat | .0370203 .0370203 .4018233 515
corptax | .0261412 .0261412 .5539194 515
tau_k |
tau_k | 1 1 634
tau_h | -.1796277 -.1796277 5.34e-06 634
tau_c | .8209436 .8209436 6.1e-156 634
topstat | .3049076 .3049076 2.28e-12 507
corptax | .1455464 .1455464 .0010139 507
tau_h |
tau_h | 1 1 634
tau_c | -.0567003 -.0567003 .1538664 634
topstat | .0140263 .0140263 .7527144 507
corptax | -.0936416 -.0936416 .0350374 507
tau_c |
tau_c | 1 1 639
topstat | .2002276 .2002276 5.53e-06 507
corptax | -.0151513 -.0151513 .7336056 507
topstat |
topstat | 1 1 517
corptax | .7163097 .7163097 1.61e-82 517
corptax |
corptax | 1 1 517
