Hello! I am using panel data and I did an Instrumental Variable estimation using the -xtivreg- and random effect command.
However, when I use the command -estat endog- to perform the test for endogeneity and the command -estat firststage- , Stata says that my commands are invalid,
1-Is there a way I can overcome this issue? as my instrument variable is a time invariant variable
2- what is the optimal sample size in my case ? becz for this study very limited data is available
3- Lasty, i observed that when i use log of independent variable then the independent variable shows a positive sign and when i use change in independent variable like (current week newc - last week newc ) then it shows a negative relationship with return which is also a change in current -last week return as the results in both cases are given below... i belive tht the change in independent variable is correct and shows better reflects the real case scenario.. can you also share your expert opinion here too? am i right if i use change in newc as my independent variable instead of log?
Thank you in advance.
Code:
. xtivreg return (cnewc=epi) lnpop i.week i.ccode, re vce(cluster ccode)
note: 16.ccode omitted because of collinearity
G2SLS random-effects IV regression Number of obs = 80
Group variable: ccode Number of groups = 16
R-sq: Obs per group:
within = 0.8016 min = 5
between = 1.0000 avg = 5.0
overall = 0.8051 max = 5
Wald chi2(3) = 301.38
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
(Std. Err. adjusted for 16 clusters in ccode)
------------------------------------------------------------------------------
| Robust
return | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
cnewc | -.0012327 .0000612 -20.14 0.000 -.0013526 -.0011127
lnpop | .0002761 .0002695 1.02 0.305 -.000252 .0008042
|
week |
2 | -.1428736 .0086981 -16.43 0.000 -.1599216 -.1258256
3 | -.0287474 .0200269 -1.44 0.151 -.0679994 .0105046
4 | .0723392 .0148804 4.86 0.000 .0431742 .1015042
5 | .0229504 .0094713 2.42 0.015 .004387 .0415138
|
ccode |
2 | -.0249133 .0008014 -31.09 0.000 -.026484 -.0233427
3 | .0112486 .000978 11.50 0.000 .0093317 .0131656
4 | -.0177978 .0005632 -31.60 0.000 -.0189016 -.016694
5 | -.0151013 .0005151 -29.32 0.000 -.0161108 -.0140917
6 | -.0289862 .0006398 -45.30 0.000 -.0302402 -.0277322
7 | -.0055123 .000614 -8.98 0.000 -.0067158 -.0043089
8 | -.0091792 .0008457 -10.85 0.000 -.0108368 -.0075216
9 | .0028841 .0007202 4.00 0.000 .0014725 .0042958
10 | -.0100807 .001121 -8.99 0.000 -.0122778 -.0078835
11 | -.0229465 .0003582 -64.07 0.000 -.0236485 -.0222445
12 | -.0085706 .0010427 -8.22 0.000 -.0106143 -.0065268
13 | .0126779 .0008793 14.42 0.000 .0109544 .0144013
14 | -.0110761 .0006022 -18.39 0.000 -.0122564 -.0098958
15 | -.0067423 .0000562 -120.03 0.000 -.0068524 -.0066322
16 | 0 (omitted)
|
_cons | 0 (omitted)
-------------+----------------------------------------------------------------
sigma_u | 0
sigma_e | .28572576
rho | 0 (fraction of variance due to u_i)
------------------------------------------------------------------------------
Instrumented: cnewc
Instruments: lnpop 2.week 3.week 4.week 5.week 2.ccode 3.ccode 4.ccode
5.ccode 6.ccode 7.ccode 8.ccode 9.ccode 10.ccode 11.ccode
12.ccode 13.ccode 14.ccode 15.ccode epi
------------------------------------------------------------------------------Code:
. xtivreg return (lnnewc=epi) lnpop i.week i.ccode, re vce(cluster ccode)
note: 16.ccode omitted because of collinearity
G2SLS random-effects IV regression Number of obs = 80
Group variable: ccode Number of groups = 16
R-sq: Obs per group:
within = 0.8169 min = 5
between = 1.0000 avg = 5.0
overall = 0.8201 max = 5
Wald chi2(3) = 413.85
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
(Std. Err. adjusted for 16 clusters in ccode)
------------------------------------------------------------------------------
| Robust
return | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
lnnewc | .0060875 .0020355 2.99 0.003 .002098 .0100771
lnpop | .0049055 .000267 18.37 0.000 .0043821 .0054289
|
week |
2 | -.1618011 .0098249 -16.47 0.000 -.1810576 -.1425446
3 | -.0477755 .0165408 -2.89 0.004 -.0801948 -.0153561
4 | .0528445 .0151661 3.48 0.000 .0231195 .0825695
5 | 0 (omitted)
|
ccode |
2 | -.004538 .0014151 -3.21 0.001 -.0073116 -.0017644
3 | .0365746 .0018804 19.45 0.000 .0328892 .0402601
4 | -.0165036 .0033607 -4.91 0.000 -.0230903 -.0099168
5 | -.0157174 .0036757 -4.28 0.000 -.0229216 -.0085132
6 | -.0356899 .0065512 -5.45 0.000 -.0485299 -.0228498
7 | .0103665 .0011737 8.83 0.000 .0080661 .012667
8 | .0010767 .0022675 0.47 0.635 -.0033675 .0055209
9 | .0153335 .0006886 22.27 0.000 .0139839 .0166831
10 | .0121774 .0001084 112.29 0.000 .0119648 .0123899
11 | -.0267325 .0036785 -7.27 0.000 -.0339422 -.0195228
12 | .0132934 .000287 46.32 0.000 .0127309 .0138559
13 | .0254747 .0016441 15.49 0.000 .0222523 .028697
14 | -.0025233 .0011965 -2.11 0.035 -.0048685 -.0001782
15 | -.0148851 .0031011 -4.80 0.000 -.0209631 -.008807
16 | 0 (omitted)
|
_cons | -.1213726 .0072204 -16.81 0.000 -.1355243 -.1072208
-------------+----------------------------------------------------------------
sigma_u | 0
sigma_e | .04248007
rho | 0 (fraction of variance due to u_i)
------------------------------------------------------------------------------
Instrumented: lnnewc
Instruments: lnpop 2.week 3.week 4.week 5.week 2.ccode 3.ccode 4.ccode
5.ccode 6.ccode 7.ccode 8.ccode 9.ccode 10.ccode 11.ccode
12.ccode 13.ccode 14.ccode 15.ccode epi
------------------------------------------------------------------------------
0 Response to xtivreg with random effect and optimal sample size
Post a Comment