Understanding Generalised Method of Moments
Вставка
- Опубліковано 6 жов 2024
- This video tries to simplify the understanding of generalised method of moments (GMM) technique in such a manner that beginners can comprehend. The video series will contain eight other tutorials: (1) How to Estimate One-step Difference GMM; (2) How to Estimate Two-step Difference GMM; (3) How to Estimate One-step System GMM; (4) How to Estimate Two-step System GMM; (5) How to Estimate Decide between Difference or System GMM; (6) How to Interpret GMM Output; (7) How to Generate Long-run GMM Coefficients; and (8) How to Plot Year Dummies in Difference and System GMM. So, what is GMM? A generic method for estimating parameters in statistical models; Uses moment conditions that are functions of the model parameters and the data, such that their expectation is zero at the parameters' true values; it is a dynamic panel estimator. And what is a panel data? It is also called longitudinal data; a multi-dimensional data involving measurements over time; contains observations of multiple phenomena obtained over multiple time periods for the same firms, individuals, countries etc. Watch my video on “Tips to Building Panel Data” for more information.
Why use GMM? It controls for: endogeneity of the lagged dependent variable in a dynamic panel model - when there is correlation between the explanatory variable and the error term in a model; omitted variables bias; unobserved panel heterogeneity; and measurement errors. How do you decide between performing the difference or system GMM? Rule-of-thumb given by Bond (2001): (1) The dynamic model should be initially estimated by pooled OLS and the LSDV approach (i.e., using the ‘within’ or fixed effects approach); (2) The pooled OLS estimate for ɸ should be considered an upper-bound estimate, while the corresponding fixed effects estimate should be considered a lower-bound estimate; (3) If the difference GMM estimate obtained is close to or below the fixed effects estimate, this suggests that the former estimate is downward biased because of weak instrumentation and a system GMM estimator should be preferred instead. It is also advisable to use system GMM if variable exhibits a random walk (persistent).
Two GMM diagnostic tests. The first is the test for instruments validity performed using Hansen (1982) J test and Sargan (1985) test of over-identifying restrictions: tests the null hypotheses of overall validity of the instruments used. Failure to reject these null hypotheses give support to the choice of the instruments. The second test is that for autocorrelation/serial correlation of the error term. It tests the null hypothesis that the differenced error term is first and second order serially correlated. Failure to reject the null hypothesis of no second-order serial correlation implies that the original error term is serially uncorrelated and the moment conditions are correctly specified. There are some challenges to estimating GMM. They are complicated and so can easily generate invalid estimates; GMM codes can be easily manipulated to yield different results; Does not account for cross-sectional dependence (CSD); Does not account for structural breaks; Not advisable for panel with very long time series (use PMG, MG and DFE estimators); Susceptible to variables listed in the IV set; Too many instruments weaken the Sargan/Hansen test and yield implausible p-values; Results are biased if instruments outnumber individual units in the panel; The problem of how many is “too many” instruments. Monte Carlo simulation evidence suggests that cutting the number of over-identifying instruments in half can reduce the bias by 40%. References used for this video tutorial are from: Arellano & Bond (1991): Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment. Review of Economic Studies Limited. 58(1) 277-297; Arellano & Bover (1995): Another Look at the Instrumental Variable Estimation of Error-Components Models. Journal of Econometrics 68 (1): 29 - 51; Blundell & Bond (1998): Initial Conditions and Moment Restrictions in Dynamic Panel Data Models. Journal of Econometrics 87: 115-144; Roodman D. (2009): How To Do xtabond2: An Introduction to “Difference” and “System” GMM in Stata; and Roodman D. (2014): Xtabond2: Stata Module to Extend xtabond Dynamic Panel Data Estimator. Statistical Software Components.
Follow up with dofiles, soft-notes and updates from CrunchEconometrix:
Website: cruncheconometr...
Blog: cruncheconomet...
Forum: cruncheconometr...
Facebook: / cruncheconometrix
UA-cam Custom URL: / cruncheconometrix
Twitter: / crunchmetrix
Reddit: / crunchmetrix
Stata Videos Playlist: • (Stata13):Estimate and...
EViews Videos Playlist: • (EViews10):Interpret V...
Hi beloved guest/subscriber, you have discovered my amazing UA-cam Channel tailored specifically for you and other beginners and intermediate users. Please do not keep me to yourself (lol). Kindly share my videos and links with your students, colleagues and academic community so that they too can SUBSCRIBE and learn with ease….and for the global community to be aware that applied econometrics can be simplified. My teaching approach is very practical. I adopt a do-as-I-do style. Many thanks to those who have supported me by telling others. Once again, CrunchEconometrix loves to teach, support my Channel with your subscription, your likes, feedbacks and sharing my videos with your cohorts. Follow me on Facebook, Twitter and Reddit. Love you all, greatly!!!
dear Bosede Ngozi Adeleye if we have the p-value of J statistique higher than 5% how do we interprete it ?
>5% = normality test passed.
Null hypothesis: data is normal
If p-value
Correct ✔️
This is most summarized presentation on GMM. After watching this video, it was much easier to follow Roodman (2009). Thank you!
Thanks Ritika, I am also glad that you find the content very helpful. May I know from where (location) you are reaching me?
I love you, you are saving my grade in the panel data course in my masters! Sending you love from Germany
Hahahaha, good to hear :). Wishing you the best in your exams. Much love from Nigeria!
Oh Lord! Thanks for providing us with a strong and willing academician like Dr. Ngoze. This videos and the rest of your videos are gems. We are learning so much from you that you can imagine. Your videos are always well detailed. You make the technical literature on Econometrics very comprehensible to us the newbies in Econometrics. We are very grateful to you madam, and someday I look forward to meeting with you and thank you in person. I'm a West African myself trying to build a career in Econometrics. thanks a lot.
Hi Kabineh, thanks for the positive feedback. Deeply appreciated! I'm glad to hear that my videos are helpful to a number of students and researchers. I'll keep doing my best to contribute my little to the global academic community. Please may I know from where (location) you are reaching me?
Hi! greetings from Mexico, Im an economist student.
Thanks a lot!! the information is to easy to understand, your video helps me to develop my investigation for econometrics class
Thanks for the encouraging feedback, Cristina. Deeply appreciated! Please may I know from where (location) you are reaching me?
Thank you very much. Your concise presentation and reference papers greatly helped me in catching up working with other colleagues in the office.
Good to hear, Eldo!
One of my favorite youtube videos! Thanks Dr. Adeleye!
Thanks so much for your encouraging words, Tallys! 🥰🙏💖
Hi dear ma'am, thank you for your great lessons, i would like to know:
1. How to check the endogeneity of variables?
2. Could we take the lag of control variable in gmmstyle instruments set if it is endogenous
3. which test is required before applying GMM?
strongly looking forward to have your response on it
Hi Farhan, I have created 9 GMM videos for simplicity of the technique. You may want to refer to Roodman (2009, 2014) for detailed understanding. Thanks.
Thank you so much for sharing your thoughts and knowledge with others... It's really appreciatable
My pleasure, Muhd! Thanks for the encouraging feedback, deeply appreciated!
This is incredible. Thank you very much for sharing your knowledge with us. So precious 😊
Compliment is humbly taken, Sanja! May I know from where (location) you are reaching me?
Thanks 👍
I have been following your videos for about two years, and every new subject I discover in your channel is impeccable. Keep up the good work, Professor Adeleye! Greetings from Bolivia.
A million likes for your encouraging feedback, Diego. Much love to Bolivians from Nigeria!
Thank you Dr for this informative video. I am on way to watch the next video.
Glad to hear you find it helpful! 👏
simplified presentation. great
Dear our instructor I would like to say thank you alot...after seeing your vedio repeatedly with out any third person support I have accomplished my MSc thesis analysis and presented in a good way.
On the defense the examiners have commented me to add endogenity test on the SVAR test, therefore this, is to request your lovely support how I can conduct the test and is important to conduct endogenity test in SVAR?
Your immediate response is highly appreciated
Hi there, thanks for your encouraging feedback. Deeply appreciated! 🥰
Unfortunately, I'm not versed on SVAR that's why I have no videos on the technique. I'd advise you to check out other online resources for more information.
Okay, but you have alot of vedios on Vector autoregressive analysis among them I have worked on structural vector autoregressive...so am sure you will have...
I want only if u show me how to conduct exogenity and endogenity test for time series data using eviews
I have no video on that. You may want to check out other online resources.
appreciated! kindly upload the video on time series as well.
Hi Sheena, video on panel VAR-GMM is uploaded to my Teachable paid platform cruncheconometrix.teachable.com. A one-time payment of $200 grants access to all videos published in the School in addition to access to free datasets, Stata dofiles and reference articles.
Thank you for your help, professor!
You are welcome!
You have been a great teacher so far
Thanks Eljay, for the positive feedback. Deeply appreciated!
Great Video, and thanks for sharing the references along with the lecture.
Glad you enjoyed it, Yogesh!
Why use GMM? It controls for endogeneity of the lagged dependent variable in a dynamic panel model - when there is a correlation between the explanatory variable and the error term in a model; omitted variables bias; unobserved panel heterogeneity; and measurement errors. How do you decide between performing the difference or system GMM? Rule-of-thumb given by Bond (2001): (1) The dynamic model should be initially estimated by pooled OLS and the LSDV approach (i.e., using the ‘within’ or fixed effects approach); (2) The pooled OLS estimate for ɸ should be considered an upper-bound estimate, while the corresponding fixed effects estimate should be considered a lower-bound estimate; (3) If the difference GMM estimate obtained is close to or below the fixed effects estimate, this suggests that the former estimate is downward biased because of weak instrumentation and a system GMM estimator should be preferred instead. It is also advisable to use system GMM if the variable exhibits a random walk (persistent). Can you please tell me from which videos should I watch this?? Please
I suggest that you watch all my GMM videos for in-depth knowledge. Thanks 😊
Thank you for the helpful video 👍🏻👍🏻👍🏻👍🏻
You are welcome, Abu! 🙏🥰
This video is extremely helpful! My advisor and committee suggested I use Arellano-Bond and/or GMM but I wasn't very familiar with any version of the technique.
Thanks for the encouraging feedback, Dark Choco. Deeply appreciated! Please may I know from where (location) you are reaching me?
@@CrunchEconometrix I'm in the US. I don't feel comfortable disclosing more specific geographic details.
I don't need specifics. Just to know the reach of my Channel. No big deal.
I like this better than indians teacher, the pronunciation are so good!
Thanks, Adi for the encouraging feedback. Deeply appreciated!..and Indian teachers are extremely good too! 🙏 ❤️
You have been very helpful for my thesis! I appreciate all of the information you have shared :). I also would like to ask how do we determine the right instruments? Do we include all of our control variables as instruments? Or do we need to test for endogeneity? If so how do we do this exactly?
I will advise that you read Roodman (2009, 2014) for more insights into the GMM model. See references listed at the end of the video.
Another great video! Thank you for uploading it!
Thanks, Pedro. Glad to hear it's helping!
From Oklahoma, USA. Great video! Thank you for uploading.
Thanks for the encouraging feedback, Jack. Deeply appreciated!🙏
Thank you so much for sharing. This video was extremely useful for me :)
Glad it was helpful, Harvie!
Your videos have been wonderful and helpful. Thanks a lot for the contribution
Thanks Abdul for the encouraging words. Kindly help to share in order to reach more people. They need to know...massallah!!!
@@CrunchEconometrix I am sure inviting a lot of people to this page and the blog.
great explanation, big help thanks!
U're welcome, Dan... glad you found it helpful! 💝
Your videos are amazing! Thank you very much!!!
Thanks, Sergio for the encouraging feedback.... deeply appreciated! 🙏
@@CrunchEconometrix Thank you professor! Because of your videos I'm starting to understand GMM! Even so, I still have a doubt. Are GMM models always specified in log-log form? Would it be wrong to specify a GMM model where the dependent variable is an index (percentage) and the regressors are logarithms?
Hugs from Brazil! 🙂
thank You Prof.. waiting for next lecture......
U're welcome, Hassaan! Please share with your friends and students too! 💕 😊
I hope further clarification. I think the longitudinal and panel data are two different things. Longitudinal data is when the data is collected at two or more than two points in time while panel data contains cross-sectional data and time series data
Hi Abdul, they are used interchangeably. Do further findings online.
First of all I would like to thank you so much for your efforts in creating videos and uploading them on UA-cam. The videos you have uploaded on panel data estimation are very much helpful for learners like me.
However, I have a few doubts about the estimation of static and dynamic panel data. such as listed below;
1. Can we use year or time dummies in the fixed effects model?
2. In one-step and two-step differences GMM, how can we get constant values?
Thank you once again Madam.
Hi Surendra, thanks for the encouraging feedback on my videos. Deeply appreciated. (1) yes; (2) xtabond2 is a User-written code and getting the CONSTANT will depend on the xtabond2 syntax you are using.
Dear Bosede, in this video you have discussed about the range of Hansen test but what should be the range for Sargan test? if someone run the one step. Thank you :-)
I constrain myself to the Hansen stat for reasons given in the video. Thanks.
@@CrunchEconometrix thank you.
This is amazing!!!!! Thank you very much!!!!
You're very welcome, Ksenia!
Thanks for the helpful videos! Subscription all the way from Seoul South Korea :)
Hi Education, thanks for the positive feedback and remarks on my UA-cam videos....I'm honored by your subscription too. Please share my videos with your friends and academic community in Seoul! 😀
Can u please share a UA-cam video on how to how to do moderation or mediation in System GMM method
You may want to check out other online resources. I currently don't have such video at the moment. Thanks
A great lecture dear Professor Dr.
I have two questions ma'am professor....
1- How to test for endogeneity in the panel data in Stata?
2- If N
Hi Rain-Walker, I mentioned in the video that GMM corrects for endogeneity that is the essence of GMM in the first instance. I'd suggest you read those references on GMM as indicated in my video for better understanding. If N
Great work. Keep on the good work
Hi Edwin, thanks for the encouraging feedback. Deeply appreciated! Please may I know from where (location) you are reaching me?
very well done, thanks
Thanks, Steven for your encouraging words. Deeply appreciated! 🥰🙏
long time don't see your video thank's for you effort
Thanks pal, working behind the scenes. Will upload soon😄
Thank you dear maam for your wonderful lessons, could you pls shed some light on how to take the log of negative value. in my data set there is some negative value. looking forward to have your insights. thank you.
Hi Farhan, I know that some researchers do but I don't take log of negative values to prevent losing observations. You may want to check out other online resources on how to get around this. Thanks.
Thank you for making this video
U're welcome Ibnu...kindly share my Channel link with your students and academic networks. Love you from Nigeria! 💕
Thanks a lot!!! Now I can see some important details that I could´t notice before.
Hi Ivan, I'm glad to hear that you find this video helpful😊
Very Helpful❤
Please help to make me out from a great dilemma....
Is Engoneous variable can be instrumental variable in GMM analysis??
Pancha, endo variable CANNOT be used as IV.
Thank you for your effort sir! Could you explain what are N and T and why N should be less than T?
Hi Mukha, please read the articles listed at the end of the video for details on GMM. N = cross-sections. T = time dimensions.
can you please also elobrate the upper and lower bounds of hansen j statistic, and sargan test, individually and the role of both AR(1) and AR(2) in interpreting the results
Hi Ishtiya, I indicated references at the end of my videos to solidify understanding and answers to questions. Kindly go through any of them, thanks.
great presentation, you have made me like GMM
Thanks the idea, Joshua! 😀 💃 Kindly share the link to my UA-cam Channel with your friends and academic community. May I know from where (location) you are reaching me?
This video was very much helpful. Thanks
Thanks for the encouraging feedback. Deeply appreciated!
Professor, I have got better journal articles having used Sys GMM on the data with T greater than n.
Is it ok, can we use system GMM in such case..
Thanks for your acknowledgement
GMM is designed for N>T panels. Read Roodman (2009, 2014) and other supporting papers listed at the end of my GMM videos.
Another great video. Thank you Prof.
Hahahaha, thanks WW2. Just when I thought you were all still sleeping. It's 3am here in Nigeria :)
@@CrunchEconometrix Great Job done.... Thanks, Madam. when the other videos are upload.
@@muhammadilyas9677 Thanks Muhd for the encouraging words. Kindly help to share in order to reach more people because they need to know. Other videos will be uploaded in due course in sequential order as stated in the video. Hold on a bit, and you'll have them all....massallah!!!
Hello Professor, Thank you for this very rich video for us first-time researchers. I am currently performing an analysis and need to use the Generalized Moment-Based Regression (GMM) method. I wanted to know how to determine the number of lags to include in the endogenous variables.
Hi Ismaila, thanks for the encouraging feedback. Deeply appreciated! There are no strict guides to determining the number of lags in GMM. You play around with different lag lengths till your model passes the mandatory diagnostics in addition to having good number of statistically significant coefficients.
Thank you, very helping...
U're welcome Riko......may I know from where you (location) are reaching me?
Always thanks for your invaluable lectures. May I request how to construct Gravity model in Eviews.
Hi Si, unfortunately I have no idea. You may need to check other online resources. Thanks.
Hi, your videos have helped me a great deal... thanks very much. Please I’ve been having serious issues analyzing panel data. I wish you could be of help please.
Hi girl, thanks for the kind words on my videos, deeply appreciated. Sure, I'll do my best to assist anyone though not personally because that'll wear me out considering my own personal schedule and time constraints. But if you can follow my videos and practice what I teach, I'm positive you'll do just fine. I don't have all the answers because I'm still learning too. But whichever area I can guide you, I will. Thanks.
Awesome sharing! Thanks for the incredible instruction.
May I ask what does External instruments and Internal instruments mean in 5:04? How do I determine which should use be used when running STATA?
Hi Daniel, thanks for the encouraging feedback. I explained what they mean in the video. External instruments are the iv() and internal instruments are the gmm(). You decide the categorization based on theory and intuition.
@@CrunchEconometrix Thanks for your prompt reply! looking forward to your future videos :)
Thank you for a well structured and easily understandable presentation. I will like to ask, how long would a panel be to make it not appropriate to apply GMM?
Hi Matthew, thanks for the kind words on this presentation. Humbly appreciated. On your query, if N
Awesome , Can you do a video where you apply both differenced and system GMM
Thanks Jonathan, yes that's part of the topics slated as you saw the outlined tutorials. Please be patient a bit, I'll definitely get to it. I'll appreciate if you can share this with your academic community and colleagues too. They need to know! 💕 😊
Please upload video on panel data in Eview and gretl software
Regards
Hi Rohtash, thanks for your suggestions. Deeply appreciated 🙏
Awesome work. Indeed this is very useful and easy for scholars. Thanx for such efforts.
Thanks Muhd for the encouraging words. Kindly help to share in order to reach more people. They need to know...massallah!!!
Dear professor,
I've studied some papers they have used GMM even if their no. of cross sections are less than time period, can you explain that why it is so? It'll be really helpful for me. Thanks for sharing your knowledge with us❤️
Hi Zeenat, I refrain from commenting on what others do. I will only do my best to educate and teach the appropriate techniques. Kindly go through the references indicated at the end of my GMM videos on the nitty-gritty of the GMM technique. Kind regards.
@@CrunchEconometrix thank you professor
Thank you for this video. I'd like to ask is the SGMM ideal for estimating the dynamic impact when N is small (the number of cross sections is 16) and T is large (the time period is 40 years). If it's not - which method would be considered ideal? I'm estimating the dynamic impact of consolidation programs on income inequality. Thank you so very much again.
Hi Angie, panel ARDL techniques are applicable. I have videos on them. Kindly search within my Channel. Thanks.
Hello, I am very grateful for your great work. Before I start using GMM technique, I just want to test for endogeneity after having used Panels Corrected Standard errors (PCSEs). By endogeneity I mean the situation in which an explanatory variable is correlated with the error term. How can I test for endogeneity after having used Panels Corrected Standard errors (PCSEs) in STATA? I could not find any clue witrh regard to this topic. In generall, what post-estimation diagnostics tests am I supposed to do after the PCSE? Preferably in the STATA software package. Professor, I would be grateful for any guidance.
I have no idea about PSCE. You may seek other online resources.
Hi thank you for this video. Question for external instruments, if i dont have external ones, can I omit it to my code?
You need to put instruments in the iv( ). Otherwise, your code won't execute.
Hello teacher,
First of all, I would like to thank you for your videos which help us a lot. Please I have a question for you: I would like to estimate a dynamic model by applying the GMM for a number of individuals which is equal to 16 and T=10, is this feasible? If not, is there another method that I can apply for this case? Thanks in advance.
Hi Khalik, GMM is in order. Please go ahead with your analysis.
Thank you teacher
Dear Professor Adeleye,
I have three questions, will be very appreciate if you can help me.
1. Can I use dummy dependent variables in System GMM estimation? If not, what kind of other method would you suggest?
2. Why we must include year dummy variables in GMM model?
3. Another question related to panel data OLS, if I do the regression using Dynamic GLS , but all variables except the lagged independent variable are insignificant, what could be the reasons?
Thanks
Best
Jing
Hi Jing, (1) you can't use dv as depvar because yu have to take its lag as explvar. Not sure of the appropriate estimator at the moment; (2) watch my video on year dummies in the GMM series; (3) not familiar with dynamic GLS. Never had cause to use it. Hope these are helpful, thanks.
@@CrunchEconometrix Dear Professor Adeleye,
Thanks for the answer.
Maybe another question , how can I specify the model if I want to use system GMM? Maybe just use the model specified by a simple pool OLS?
Thanks!
Jing
@@jingma603 I'll refer you to any published paper that used GMM. You'll see the way the model is specified and explained.
Ma'am Videos are really helpful. can you make videos on MGARCH
Thanks for the encouraging feedback, Karan! I'll do the video once I understand the technique.
Very good!! thk u.
U're welcome! 🥰
Prof, you are good, thank you so much
Thanks for the positive feedback and remarks on my video, Darcy. Deeply appreciated! May I know from where (location) you are reaching me?
@@CrunchEconometrixyou are most welcome Prof, Zimbabwe (Masvingo Province)
Awesome! I'll appreciate if you share my codes with your friends and academic community. Thanks.
Awesome! I'll appreciate if you share my codes with your friends and academic community. Thanks.
i have subscribed and looked your econometrics videos really they are constructive but for the time being i ask two questions on pane data analysis
1.how can i decide between static and dynamic panel?here I have watched your video which deals on selection on sgmm and dgmm but first how can i decide wether my model is dynamic or static panel
2.which tests are mandatory in panel data analysis? may God bells you from Ethiopia
Hi Sisayne, I am humbled by your kind remarks and your subscription...deeply appreciated. Engaging a static or dynamic analysis is at the discretion of the researcher. You may need to read related studies for more constructive reasons. My panel data videos contain post-estimation tests, kindly watch them. Thanks.
Hello Professor. Thank you for your great videos. I am working on an unbalanced panel data with possible endogenous regressors. However my N
Hi Harshali, thanks for the positive feedback on my videos. Deeply appreciated! 💕 If N
@@CrunchEconometrix I am from India.
@@harshalidamle7161 Awesome! 💕 I have lots of Indians in my CrunchEconometrix Facebook Community...keep watching, keep sharing.
@@CrunchEconometrix I also have a case where N
Hello from France and thanks a lot for the video and your work. I want to know if we don't
need to test for unit roots if we want to use this estimator.
Hi Sabir, unit root test not required.
Thanks
You're welcome!🙏
Must follow Roodman(2009), she explained the similar way, thanks for explaining this paper, can You explain on panel var and structural var using Stata or R. THANKS IN ADVANCE
Hi Shakti, I'll do my best to make those videos once I fully understand the techniques. Please may I know from where (location) you are reaching me?
Great... It was helpful
Keep watching, Isaac...thanks for the encouraging feedback. Appreciated!
Hello dear thank you for your excellent contribution.
In calculating xtabond2 I am facing the error by not getting the AR(2) value...how ever the value of AR(1) is quite significant with no error of serial correlation. I am using quarterly data and my t>n. So how can I get the value of AR(1) and AR(2) at the same time. If you have any relevant videos, please post the link so that I can see it. Thank You..
Hi Dr. Faisal, GMM is only applicable to N>T panel structure.
Good every ma. How was your day?
I was watching your video on GMM and you made mention of some conditions that must be satisfied in order to use it. My question is on the N > T condition.
I want to do a cross country study (3 countries) with a total of 220 companies for 11 years. But someone said I can't use GMM because my cross section isn't just the companies but also the countries and the number of countries is less than the number of years.
What is your opinion? If truely I can't use GMM, what other dynamic panel estimator can I use?
Hi Edosa, my day is gr8, thanks. The GMM procedure is applicable because your cs are the 220 companies NOT the 3 countries.
How are you Dr. Thank you very much for the videos. Please, I have a question. I would like to estimate a dynamic model by applying the GMM on eviews but the issue is on the part of presenting results, there is an error messsage saying number of instruments are greater than number of observations. So Dr, should I continue using GMM or there is something I should do?
Hi Takunda, due to the command-driven interface, I find Stata more robust for panel data analysis. I will advise you to post this to any EViews platform for constructive feedback. Thanks
Minute 11.27 AR1 not necessarily to be rejected especially when you do Difference GMM. Differenced values would always correlated with first order error term. Failure to reject AR2 test is a must.
Very true, Ahmad, failure to reject AR(2) more important which I emphasised. Thanks for watching, kindly share with others too! 💕 😊
Hi! so it's actually fine to accept the null hypothesis of no first-order serial correlation, isn't it? And what about system GMM?
@@caosang8079 null hypothesis of AR2 must not be rejected to indicate no serial correlation of the second order. This is an important condition to make GMM (system or difference) estimators consistent.
@@abwayman I understand that part about AR(2). But what I'm concerned about is AR(1). Is it fine to accept no serial correlation in AR(1) because I got the insignificant result regarding AR(1) (i.e. p-value > 0.05)
@@caosang8079 oh yes, if your ar1 is not significant, that is what we wanted actually. The issue is if it's significant. But significant ar1 is fine too, for the reason i mentioned in my first comment. Ar2 though must not significant.
Ma Thanks for the Video on GMM
Am working on a paper on GMM likewise.
The one step Gmm are well defined. But I noticed that the diagnostic of sargen and Ar bond did not meet the requirements based on decision rule . Like wise in two way system Gmm.
What do you suggest could be done.
Hi Jo, have you watched any of my GMM videos because i explained what to do. Also, I explained in the foundational video that Hansen J is more robust than Sargen stat. I advise you watch these videos for clarity. thanks.
Hi Ma'am . Can you plz guide me about iterated gmm estimation? Urgently needed.
Hi Sobiya, I don't quite get your query.
In a study, if i conduct both fixed effect model regression (one independent variable turned out significant) and gmm regression (two significant variables turned out significant -including the FE model's variable).
What should i include in the paper for panel data analysis?
Or should i include both
Anik, using FE and GMM may depend on your empirical approach and if that addresses your research objectives. So, I will leave you to decide. Thanks
Thanks for the video, those are helpful! I was wondering if a SYS-GMM estimated could be valid even if the Diff-GMM doesn’t show a significant impact of the lag dependent variable.
Thanks, William for the positive feedback. Deeply appreciated. Yes, that can make a case for using the sys-GMM is the study intent is to show the persistency of the depvar. May I know from where (location) you are reaching me?
@@CrunchEconometrix thank you for you quick response, I'm from Canada :).
@@williambriggs9817 Awesome!!! I'll appreciate if you can share the link to my YT Channel with your students and colleagues in Canada for awareness...thanks!
Dear prof. Adeleye, I want to thank you for giving interesting and insightful lectures on difficult topics. I would like to ask a question related to the challenges connected with use of GMM estimator. It is said that if a dynamic panel data contains very long time series, the use of an estimator alternative to GMM is more advisable. My dynamic panel data contains economic data on more than 200 banks from 12 countries during 11 years. I intend to use country fixed effects and time fixed effects in the regression. As my regression contains one lagged dependent variable, would you recommend using GMM estimator or an alternative?
GMM technique is designed for short panel analysis when N>T. Kindly watch my video on "Tips to building panel data".
Hello Prof. Thanks for this insight. Please can one use GMM to estimate panel data without including the lag of the DV?
Thanks for the encouraging feedback, Horvey. Deeply appreciated! GMM is a dynamic estimator which require the inclusion of the lagged depvar.
@@CrunchEconometrix Thanks Prof
Hello dear Professor, thanks for your videos on GMM, it helps a lot, but I also would like to ask that can we apply GMM model to time series data?
Yes, its applicable. There are some Stata syntax to that (use the Help file) but I have never done that approach before. You may also seek other online resources.
Hello Professor, thank you very much for you video. However, sorry about this basic question I'm about to ask and my bad English. Actually I still can't quite understand about number of instruments. Can you help me describe what is the exact meaning of number of instruments in GMM? thank you
Hi Stefany, thanks for the positive feedback. On "instruments", I will refer you to read some of the literature indicated at the end of the video for better understanding. Thanks.
Hello Dr. Addeleye,
Thank you for making econometrics look easy for students. I really appreciate your videos as they have served my understanding very well.
However, i would like to know in the event that T>N for a dynamic panel case, what estimation techniques are best? Also, can i use the Fully modified OLS and Dynamic OLS to estimate a dynamic panel model?
kindly awaiting your response.
With regards.
Thanks Talatu, I am also glad that you find the content very helpful. Kindly watch my videos on panel ARDL as that answers your query but unfortunately cannot say if FMOLS is applicable. May I know from where (location) you are reaching me?
@@CrunchEconometrix From Ghana.
I have watched the video you recommended. I got my answer. thank you.
@@talatujalloh8720 Good to hear that. Kindly share the link to my UA-cam Channel with your friends and academic community in Ghana 🇬🇭 for awareness. They'll learn some useful tips and hints too. Thanks!
Can you please explain how number of instruments are calculated in system GMM. I tried to search this throughout your lectures and from other sources as well but couldn't find it anywhere.
Ammar, read Roodman (2009, 2014).
Please what should be the range for the coefficient of the lagged dependent variable in the dynamic model given that the coefficient is significant at 5%?
Kofi, I'm not aware of any range of values for the coefficient of the lagged depvar.
@@CrunchEconometrix Thank you very much
thank you
You are welcome 🥰🙏
Hi Proffessor.Thank very much for your outstanding video on GMM. However, as a beginner, I am using system GMM, but the results on Hansen & Sargan kept giving significant values, what do u recommend?
Hi Yassin, several simulations can be done. Change regressors, instruments sets etc. Advisable to watch the rest videos in my GMM series to know how to approach this. Thanks.
Can I use the GMM for an Annual Time Series data that has been converted to Quarterly data ?
Hi Alfred, yes you can. But this video is based on GMM using a panel data NOT time series.
Prof, thanks very much , I'm coming once more to the SGMM technique. I have a question (which you addressed but still have some doubts). I have have a Hansen of 1. When i used the collpase option as you adviced, the p-value still remains 1. I have changed both the external and internal instruments but nothing changes. So , I was wondering how can i Justify a Hansen's P-value of 1 (during a presentation).
Asante, you can't if someone on the panel knows about GMM. You may need to change your regressors and run several simulations.
@@CrunchEconometrix thanks my beloved professor
Hello dear. What is the best model to be used when the number of observations are small?
Charles, I have no direct response to this. Minimum sample/observations size for regression analysis is 30.
Great video! Are there any assumptions necessary for using a GMM estimator that cannot be empirically tested? (Are there any assumptions that would require a qualitative argument?)
Hi Noah, honestly not to my knowledge...and thanks for the positive feedback. Deeply appreciated!
@@CrunchEconometrix Thank you! This is good news haha
Hi Dr Ngozi, i wish to know if the abscence of the year dummies biases our gmm results, and if yes, do you have a video explaining how we Can generate It in stata?
Hi Asante, not really though it accounts for variation in the depvar across the years. Scan through the literature and you'll see that some don't include it during estimation. You generate year dummies with:
tab year, gen(d)
You can purchase my dofile to get other codes. Thanks.
hi, i have difficulty in using xtabond2 in stata. do you understand how to combine levels or lag, difference, and orthogonal? very happy if you can help me :)
Viary, I have 9 detailed practical videos on GMM estimation. Find the time to watch them as they address all your queries. Thanks.
Hi
Thank you for your incredible channel.
What about random effect models?
My panel data is random effect and one of my explanatory variables is not station at level and Wooldridge test showed my panel data has autocorrelation problem. Shoud i use dynamic panel model? My panel data contains 7 cross-sections and daily data for more than 8 years (about 23000 cells)
Thank yo
Hi Hossein, kindly watch my video on "Fixed and Random Effects Models". Thanks.
What is the best model for a small sample panel with T>N , the heterogenous slops and cross section dependent. thanks.
Have at least 30 years observations and deploy any panel ARDL technique.
@@CrunchEconometrix i have N=7 and T=21 and cant increase it what should i do in your opinion?
Sarah, transform the yearly data into quarterly that gives you 84 quarterly observations.
@@CrunchEconometrix I appreciate your suggestions, and I like to thank you for your amazing tutorial videos' in econometrics.
Thank you very much for your video, it was very helpful! I have a question regarding SGMM that I was hoping you'd be able to help with if that's ok? I have run the regression 'xtabond2 gdpg lgdpg d pop educ loginv, gmm(L2.(lgdpg d pop educ loginv), collapse) twostep robust nodiffsargan orthogonal' on stata to examine the relationship between public debt and economic growth (accounting for population, investment and education) however, I have 111 instruments and so the Hargan/Sansen tests suggest the model is weakened by many instruments. Do you know what the problem is/how to solve it?
Elliot, your specification bloated your instruments. Reduce the variables in the gmm( ) set.
Thank u so much 😊
U're welcome, Zorni...may I know from where (location) you are reaching me?