How to Estimate Models with PCSE Technique: Pre-Estimations
HTML-код
- Опубликовано: 8 сен 2024
- What is Panel-Corrected Standard Errors (PCSE) Technique?
1) Controls for cross-sectional dependence, autocorrelation and heteroscedasticity.
2) Applicable to N less than T panel data structure - when the number of cross-sections is LESS than the time dimensions.
3) PCSE is a static panel data technique…suitable for long-run analysis.
This video replicates model [1] in Adeleye et al (2022) “Does Globalization and Energy Usage Influence Carbon Emissions in South Asia? An Empirical Revisit of the Debate”. Environmental Science and Pollution Research doi.org/10.100...
Article available on my RG profile. Kindly make a request and I will send it to you.
Baseline model: 𝐥𝐧〖〖𝑪𝑶〗_𝟐〗_𝒊𝒕 = 𝜷_𝟎 + 𝜷_𝟏 〖𝐥𝐧𝑷𝑪〗_𝒊𝒕 + 𝜷_𝟐 〖𝑷𝑮𝑹〗_𝒊𝒕 + 𝜷_𝟑 〖𝑹𝑬𝑵〗_𝒊𝒕 + 𝜷_𝟒 〖𝑹𝑸〗_𝒊𝒕 + 𝜺_𝒊𝒕 [1]
Carbon emissions expressed as a linear function of per capita GDP, population growth, renewable energy and regulatory quality
PCSE Analytical Procedure:
1) Specify the model
2) Check for multicollinearity (perform correlation analysis)
3) Test for cross-sectional dependence* (Pesaran CD Test)
4) Perform unit root test (CIPS, CADF)
5) Test for cointegration (Westerlund Test)
6) Estimate the model (using PCSE)
CrunchEconometrix videos should be supported by relevant readings from econometrics textbooks, journal articles and other resources to properly harness the simplicity of the video tutorials.
Suggested References:
1) Adeleye, B. N., Akam, D., Inuwa, N., James, T. H., Basila, D. (2022) “Does Globalization and Energy Usage Influence Carbon Emissions in South Asia? An Empirical Revisit of the Debate”. Environmental Science and Pollution Research doi.org/10.100...
2) Adeleye, B. N., Akam, D., Inuwa, N., Olarinde, M., Okafor, V., Ogunrinola, I., Adekola, P. (2021). “Investigating growth-energy-emissions trilemma in South Asia”. International Journal of Energy Economics and Policy, doi.org/10.324...
3) Adeleye, B. N., Bengana, Boukhalkhal, Shafiq, and Abdulkareem (2022). “Does human capital tilt the population-economic growth dynamics? Evidence from Middle East and North African Countries”. Social Indicators Research doi.org/10.100...
4) Adeleye, B. N., Ola-David, O., Jamal, A., Arumugam, S., (2021) “Moderation analysis on tourism-growth-emissions nexus in South Asia”. Journal of Policy Research in Tourism, Leisure & Events (Special Issue) doi.org/10.108...
Hello, i'm from Indonesia and thank u so much for making this video. It helps me so much
You're so welcome!
Hello Dr. Adeleye,
Thank you so much for this incredibly informative and practical video. The knowledge you shared has been instrumental in helping me overcome several challenges. I truly appreciate the clarity and depth of your explanations.
Beside that, I have one question: Is it necessary for all variables to be first-order stationary I(1), or can they be stationary at different orders, such as level I(0) and second-order I(2)?
Thanks again for your help!
Glad it was helpful!...the dependent variable should be I(1).
Thank you for this video, please can you make one teaching about Nonlinear ARDL model?
Hi Sebastian, I'm still gathering resources on the technique. Once I'm confident enough, I'll create the videos. Thanks for suggesting...deeply appreciated! 🙏
Hello, thank you for this informative video. I would like to ask if cross-section dependence is present in some of the variables and not present in others, I can employ this approach. Also supposed if CIPS shows I(0) for all variables can I follow this approach??
Yes, absolutely. Even if CS is present in only one variable, go ahead and deploy the PCSE technique.
@@CrunchEconometrix Thank you for the feedback.
Great video. Thank you Madam.
Please, l used PCSE for a N=50, T=20 panel and l had a good outcome. All the other checks as stated in your video were successfully done. Does this still mean that my model is not correct since my panel is N>T?
Hi Dennis, thanks for your encouraging feedback. Deeply appreciated 🙏. Your model may pass due to several factors but does not negate the fact that the technique is suited for N
Fantastic video Dr Adeleye, thankyou very much. However I have a question: I read that the Im, Pesaran, Shin test is a 1st generation test, not 2nd generation. I understand that if you subtract the cross-sectional means is makes the Im, Pesaran, Shin test more robust to cross-sectional dependence, but isn't it better to use a true 2nd generation test such as Pesaran CIPS/CADF test using the "pescadf" or "xtcips" commands?
Thanks for your contributions. Stata inbuilt IPS is suitable for models with CSD, which is why the algorithm has the option to control for "cross-sectional means."
Up you Dr Ngozi Adeleye more grace to do more for this generation
Thanks so much, Mum...deeply appreciated! 💖🙏
Very nice video Professor, please for the application of this model, is it only the dependent variable that is necessarily I(1) or necessarily all the variables of the model including the explanatory variables?
Thank you for your kind words! In this model, it is typically the dependent variable that is considered to be I(1), but it is always a good idea to check the stationarity of all variables in your model, including the explanatory variables.
Thanks very much for the clarity in the explanation Dr. I have three questions please. 1. I tried testing for the CSD but it only works when I delete some of my control variables. When I insert all my control variables, the feedback I get from stata is “insufficient observations” what should I do Dr?
2. I used this model and my supervisor raised the issue of endogeneity concerns, does this model deal with that?
3. Lastly, I texted for endogeneity using the IV regress and my variables were exogenous, do I have to report the results for both Pcse and IV regress or I should only report the results for the pcse and add the test of endogeneity to it? Thanks Dr. God bless you
1) could be that some of the regressors have insufficient observations.
2) it doesn't correct endogeneity.
3) that depends on what your Supervisor wants.
Thank you very much. I have a question:
How can we control panel heteroscedasticity, serial correlation, cross section dependence and endogeneity problems?
Using the PCSE controls for these.
Hello, Thank you for video. Can I ask you why variables REN and PGR are not logarithmic values? In various studies, I encountered the transformation of % values into logarithms
@BlazejS transforming into natural log is at the discretion of the researcher. Though, not advisable to transform "rate" variables.
According to "Which panel data estimator should I use?" by (Reed & Ye, 2011). PCSE provides more efficient results when N>T. however in few other papers its opposite. So according to Reed & Ye, 2011 can we use PCSE when N >T?. kindly provide your feed back. thanks in advance
Hi Muhd, you can use the reference you have cited to support the approach you intend to follow. You may also want to reference the articles I cited in creating this video. Thanks.
Hello I need your help with estimating TFP using Cobb-Douglas function
Hi Pacha, not sure if I have an idea on how to do that.
Very helpful. Can we use this estimation technique for unbalanced panel data?
Yes, you can.
Well done crunch Queen. beautiful explanation. But Dr is Im et al(2003) and Pesaran(2007) both second generation unit root test?. I knew Im, Pesaran, and Shin (2003)(IPS) is first generation and cross sectional augmented IPS(CIPS) of Pesaran (2007) is second generation.
Thanks for your contribution, Sir. Having controlled for cross-sectional means, the IPS becomes a 2nd generation URT.
Hello, can I ask for a question: I cannot perform westerlund test, when I submit, it appears "No observations". How can I fix this?
Thanks
Hai, that may be due to one of your variables not having a sufficient number of observations to run the test. I'll suggest you scrutinize your data and drop that variable.
Hello Dr Adeyele I have a question: is pcse applicable when t
From the Stata HELP menu, the algorithm is designed for T > N panel structure.
@@CrunchEconometrix I see Dr... Can you recommend any statistic test that can be used for short-run estimates? Such as when
T < N?
I don't understand what you mean?
@@CrunchEconometrix hi Dr. let me rephrase my question.. currently we're running a panel regression for short panel data wherein T is smaller than N (T
Check out Stata HELP menu for "pscc" technique. Suitable for N > T panels.
How can I estimate the parameters using a fixed effects model under the PCSE procedure
Hi Tarek, not sure how to do this. You may want to check out other online resources. Thanks
@CrunchEconometrix thank you, But how can we ensure that the problems are corrected if the PCSE model is used?
Hi Tarek, type "help xtpcse" into the COMMAND WINDOW and read more about the technique to clear all doubts.
@@CrunchEconometrix Okay doctor, but is PCSE an independent model? Meaning, when interpreting the results, do we indicate that the estimation was done using the PCSE model, or do we mention what?
Tarek, kindly watch the PCSE videos again. I gave an informative presentation. If in doubt, check the Stata HELP menu for more information about the technique.