Dear Anuradha Saikia ,

In the first place, for better or for worse, take into account that the Hausman test is of limited value for many referees. They demand you using fixed-effects techniques anyway in order to ensure consistency. The reason is that

"A final word of caution. In reading empirical work, you may find that some authors decide on FE versus RE estimation based on whether the ai are properly viewed as parameters to estimate or as random variables. Such considerations are usually wrongheaded. In this chapter, we have treated the ai as random variables in the unobserved effects model (14.7), regardless of how we decide to estimate the j . As we have emphasized, the key issue that determines whether we use FE or RE is whether we can plausibly assume ai is uncorrelated with all xitj. Nevertheless, in some applications of panel data methods, we cannot treat our sample as a random sample from a large population, especially when the unit of observation is a large geographical unit (say, states or provinces). Then, it often makes sense to think of each ai as a separate intercept to estimate for each cross-sectional unit. In this case, we use fixed effects: remember, using FE is mechanically the same as allowing a different intercept for each cross-sectional unit. Fortunately, whether or not we engage in the philosophical debate about the nature of ai , FE is almost always much more convincing than RE for policy analysis using aggregated data." (Wooldridge)

In order to deal with this problem, the most straightforward way is to implement a regression-based version of the Hausman test proposed by Wooldridge. He describes it in the more advanced "Econometrics of cross-section and panel data" and the Stata code can be found in Cameron and Trivedi "Microeconometrics using Stata".

https://stats.stackexchange.com/questions/139466/procedure-for-the-cluster-robust-hausman-test

I have myself applied it here:

https://www.cambridge.org/core/journals/journal-of-pension-economics-and-finance/article/supplementary-private-pensions-and-saving-evidence-from-spain/91248880AAF307F27CDBE448EDA830A8

I hope it helps,

José-Ignacio

Thank You José-Ignacio Antón Sir. I will look into this

1. https://www.statalist.org/forums/forum/general-stata-discussion/general/584785-confused-with-haumsna

2. https://stats.stackexchange.com/questions/65650/hausman-test-for-panel-data-fe-and-re-error-in-the-estimation-what-to-do-sta

3. https://www.stata.com/statalist/archive/2003-10/msg00029.html

Pay attention to the rank of estimators (because of the formulation of hausman test). For example, if you type hausman fe re, then please change to hausman re fe.

Thank you so much ma'am for the references Mousumi Bhattacharya .

Dear Anuradha.

Why not try using fixed effect technique to ensure consistency if Hayman test is giving you problems.

Regards

Hausman’s specification test

Hausman’s specification test can be used to test the random effects model against the fixed effects model. It indicates that the fixed and random effects models are significantly different from each other. To decide between fixed or random effects, one can run a Hausman’s test where the null hypothesis is that the preferred model is random effects Vs. the alternative the fixed effects (Green, 2008, chapter 9). It basically tests whether the unique errors (ui) are correlated with the regressors (X’s) and the null hypothesis states that, they are not. Hence, in this study Hausman’s test is used to test that the panel data model without spatial effect is a fixed-effects generalised least square model (FE-GLS model) or random-effects generalised least square model (RE-GLS model).

Reference: R. Gangai Selvi and K. Mani, 2015, Land Use Dynamics in Tamil Nadu - A Spatial Econometric Analysis, PhD Thesis submitted at Tamil Nadu Agricultural University, Coimbatore, Tamil Nadu, India

Hello Anuradha Saikia

I suggest you first conduct the Pool mean group, Fixed Effects and Random Effect. on the choice of the best model you use your Hausman’s specification test to compare FE and RE, FE and PMG, PMG and RE

Dear Anuradha Saikia.The Hausman test informs possible problems of endogeneity. A very common problems in economics that make generalized least squares (random effect estimators) biesed. In such cases, it is necessary to verify that the set of regressors is exactly the same. If not, one of the variables may be linked to endogeneity. It is very common to have more regressors in the random effect models since variables "fixed in time" are eliminated in the fixed effect models. That said, if you are confident about the selection of variables and the Hausman text indicates differences between the models, work with fixed effects.

It is also useful to test whether the fixed-effect models generate similar results to the first difference models. If this not occurs, you may need to change the model and add lag X, or even lag Y among the regressors.

If the "true model" includes only lag x in the regressors, you can use fixed or random model and the hausman test.

If the "true model" includes lag Y in the regressors, then both the fixed and random effects are biesed. When this occurs, Hausman's text may indicate that the two models are apart (or generate different results) but neither of them will be the correct result. In these cases you can use Blundell and Bond 1998.

Hello everyone.

I am working on a panel data on Asian ccountries . T=19 and N=3

The Hausman results are showing

b = consistent under Ho and Ha; obtained from xtpmg

B = inconsistent under Ha, efficient under Ho; obtained from xtpmg

Test: Ho: difference in coefficients not systematic

chi2(3) = (b-B)'[(V_b-V_B)^(-1)](b-B)

= -9.56 chi2 model fitted on these

data fails to meet the asymptotic

assumptions of the Hausman test;

see suest for a generalized test

I even tried several ways to handle this problem but every time i got the same outcome from hausman test please suggest me a possible way to sort out this problem. thanks in advance

I'm also getting the same problem

do u have any solution

This is what showing "STATA shows me the message 'chi2 model fitted on these data fails to meet the asymptotic assumptions of the Hausman test; see suest for a generalized test"