Estimation problems

[Koffi Yves YA]

Hello dear Stata netizens.
I am a PhD student in economics. The tests carried out revealed the presence of heteroscedasticity and the logTFM variable suffers from endogeneity. I present you my estimate.
My concern is whether this resulat is good or there are still other tests to perform.
I have also made dynamic panel estimates with GMM mothodes but the number of instruments is too large and exceeds the number of groups. I extended my estimate then on 32 countries but the results are not satisfactory. The number of instruments is now less than the number of groups but no variables are significant other than the delayed dependent variable.
I first present the result with fixed effects: the choice was made through the Hausman test but Khi-two is negative so I added the option "sigmalex" and do the Mundlak test. Both tests revealed that the fixed effect model is more appropriate.
Thank you

. xtivreg IDHIx logPIBH TIBCP IDEx VOIX SP CIFSPx (logTFM = logTFM), fe vce(robust) small
Fixed-effects (within) IV regression Number of obs = 104
Group variable: COUNTRY1 Number of groups = 13
R-sq: Obs per group:
within = 0.6640 min = 8
between = 0.6161 avg = 8.0
overall = 0.6126 max = 8
F( 20, 84) = 61.19
corr(u_i, Xb) = -0.4805 Prob > F = 0.0000
(Std. Err. adjusted for 13 clusters in COUNTRY1)
------------------------------------------------------------------------------
| Robust
IDHIx | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
logTFM | 1.337852 .4530697 2.95 0.004 .4368734 2.238831
logPIBH | 3.49779 2.673472 1.31 0.194 -1.818701 8.814281
TIBCP | .1181408 .0299693 3.94 0.000 .0585435 .177738
IDEx | -.0668003 .0101499 -6.58 0.000 -.0869844 -.0466161
VOIX | 3.190805 1.874817 1.70 0.092 -.537474 6.919084
SP | -1.408451 1.857606 -0.76 0.450 -5.102504 2.285603
CIFSPx | .3235224 .1181944 2.74 0.008 .08848 .5585649
_cons | -40.60913 11.56881 -3.51 0.001 -63.61497 -17.60329
-------------+----------------------------------------------------------------
sigma_u | 3.3428154
sigma_e | 1.8462451
rho | .76626122 (fraction of variance due to u_i)
------------------------------------------------------------------------------
Instrumented: logTFM

Instruments: logPIBH TIBCP IDEx VOIX SP CIFSPx logTFM

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output (fixed spacing fonts helps), and sample data using dataex. Also, use a descriptive topic description - for your issue something about endogeneity in panel data.

xtivreg2 and ivreg2 provide a wide variety of diagnostic tests for this kind of estimate. As for GMM and number of instruments, you should look at Roodman's paper on xtabond. You can restrict the number of instruments.

While many use Hausman to decide on fixed vs random effects, it also depends on whether your theory is a theory of changes over time within panels or about stable differences across panels.

You are pushing it to have 13 fixed effects plus 8 parameters with only 104 observations. The meaning of estimates with asymptotic properties when you have so few observations relative to parameters is questionable.

[Koffi Yves YA]

Thanks Phil Bromiley