Stata16: Estimate Panel Data Models using PCSE Technique (Part 2)

[@dennisbaidoo5995]

Very insightful video. Thank you Doc.

[@CrunchEconometrix]

U're welcome, Sir. I really appreciate your encouraging feedback. ❤️

[@student7678]

Dear Dr. Adeleye, Many thanks for your insightful and easy way to learn Econometrics videos. In the reference [1] article (Adeleye et al., 2022) ...Does Globa..) you formulated two turning point figures for energy-emissions and globalization-emission paradox. Would you please tell me how you did that? What were the Stata commands for plotting the turning point figures? Thank you!

[@CrunchEconometrix]

Kindly check my LinkedIn post on QUADRATIC MODELLING. I also have the resources on my RG profile.

[@student7678]

@@CrunchEconometrix Many Thanks!!!

[@student7678]

Dear Dr., I have run your Quadratic Modelling Time series data, Everything is okay but when I check for vif (Multicolinearity test) then lntr and lntr2 vif show greater than 10 which is a problem of Multicollinearity. What should I do? Thanks! . reg lnco2 lngdp lnpop lntr lntr2, ro Linear regression Number of obs = 59 F(4, 54) = 50.75 Prob > F = 0.0000 R-squared = 0.8453 Root MSE = .1296 ------------------------------------------------------------------------------ | Robust lnco2 | Coefficient std. err. t P>|t| [95% conf. interval] -------------+---------------------------------------------------------------- lngdp | .2089165 .0198416 10.53 0.000 .1691365 .2486966 lnpop | -.3623335 .1239737 -2.92 0.005 -.6108858 -.1137813 lntr | -.4646942 .1605117 -2.90 0.005 -.7865008 -.1428876 lntr2 | .0773998 .0226952 3.41 0.001 .0318987 .1229009 _cons | -4.960184 .524272 -9.46 0.000 -6.011286 -3.909082 ------------------------------------------------------------------------------ . vif Variable | VIF 1/VIF -------------+---------------------- lntr2 | 73.60 0.013587 lntr | 71.11 0.014062 lngdp | 1.57 0.636307 lnpop | 1.18 0.846000 -------------+---------------------- Mean VIF | 36.87

[@CrunchEconometrix]

Ignore the collinearity of the squared term. It is expected. Read more about this from any online reference.

[@student7678]

@@CrunchEconometrix, Many thanks for your suggestion. Would you please suggest a reference to learn more about the square term of any variable (Quadratic Model)? Thanks!

[@KatherinPinzónNaranjo]

Hello. Thank you so much for the explanation. One question that I am struggling with. I have a panel and a have to estimate a set of regressions out of it, I have carried out the hausman test and the conslusion I got is that I have to employ fixed effects, however, I have a long panel data with heteroscedasticity and autocorrelation, then one of the alternatives I have found for this issue is estimating a PSCE model, I have already done the cross sectional dependence test and this model is still the best option. My question is: For controling the unobserved characteristics that might vary across units (which in my case are developing countries) is it necessary to include country dummies?

[@CrunchEconometrix]

Yes, you can add dummy variables.

[@EdemDOUVI]

Good Morning Madame, Please, I need help. I am doing one analysis and in my data, there is a problem of multicolinearity. To deal with that, my reference article has used the partial least squares. My problem is that I was looking for that package and the command on stata a long time ago, but I didn't see it. Your help will be so much appreciated.

[@CrunchEconometrix]

Hi Edem, I've never used PLS. You may want to post on Statalist.org for more constructive feedback. Thanks.

[@luthfianarahmaa]

Very fruitful explanation. Thanks a lot, Doc! However, several things I would like to confirm: #1. What if, in previously in OLS and FE method, there is an independent variable (which is not the variable of interest) showing different sign (but not statistically significant) as we expected, could we still continue to try the PCSE method? #2. Continuing from question #1, what if after we use PCSE methods, the independent variable stated in #1 showing the right sign (aligned with my expectation), can we correctly use the results for our interpretation?

[@CrunchEconometrix]

The statistical significance OR insignificance of coefficients and using estimation techniques are not entirely mutually exclusive.

[@luthfianarahmaa]

So I could confirm that I should continue to the PCSE method and use the results, even if the OLS and FE does not show the right sign. Isn't that correct, Doc? Thank you in advance@@CrunchEconometrix

[@Abrar-Khan09]

Hi dear, while running PSCEs estimator, is it must to take the "first difference" of variables if there is stationarity problem in the data?

[@CrunchEconometrix]

Yes, Abrar. The variables must be stationary.

[@Abrar-Khan09]

@@CrunchEconometrix thank professor, i have to take "first difference" if the variables are not stationary at the level?

[@wendyaledon9110]

good day, in running for PCSE should i use the cointegrated data, that is level form, or should i use the stationarized data?

[@CrunchEconometrix]

You may want to follow what I did. My PCSE videos are well explained.Thanks.

[@EstherOgundare]

Dr Crunch Queen Space receive more grace and more strength to do more in sound health IJMN weldone Dr Ngozi❤

[@CrunchEconometrix]

Amen and amen, Mum! 🥰❤️

[@farhanfarzam4278]

Thank you so much ma'am for your great lessons. My question is based on which model (OLS, RE, FE) can we check cross sectional dependence to determine the suitable model. dear ma'am pls reply to this questions

[@CrunchEconometrix]

Once you are using a panel data, it is important to check for CSD regardless of the technique of estimation.

[@farhanfarzam4278]

@@CrunchEconometrix thank you ma'am I have checked CSD exists in the model. And I have done 2th generation unit root tests as well, some of my variables are not stationary at level, but all is stationary at first difference. if I take the first difference I lose a lot of my data. now what should I do ma'am?

[@CrunchEconometrix]

Follow the steps shown in the video to analyse your data.

[@meerrasiq1169]

Mam, please make a video on the Augmented Mean group and Cross sectional ARDL .

[@CrunchEconometrix]

Noted. Thanks for your suggestion.

[@SAHILVERMA-fc9yr]

2) Applicable to N less than T panel data structure - when the number of cross-sections is LESS than the time dimensions. can you please provide me any reference of it. thank you in advance

[@CrunchEconometrix]

Kindly refer to Stata notes on this routine. Use the HELP menu. Thanks.