(Stata13): How to Estimate One-Step Difference GMM

[@CrunchEconometrix]

Beloved guest/subscriber, you have discovered my amazing KZbin 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, likes, feedbacks and sharing my videos with your cohorts. Follow me on Facebook, Twitter and Reddit. Love you all, greatly!!! 

[@jtalvarez1]

Thank you very much for this video. Great help! :)

[@CrunchEconometrix]

@@jtalvarez1 My humble pleasure, JT...please tell others too! 💕 😊

[@jtalvarez1]

@@CrunchEconometrix Sure I will. :) May I please know how to access the do files stated in the video? (Not sure if I just missed the instructions). Thank you. :)

[@CrunchEconometrix]

@@jtalvarez1 Google account and a Chrome browser is all you need for easy access and download.

[@jtalvarez1]

I got it. Thank you so much! Will be recommending your channel to other graduate students. :)

[@ayodejinajeemiziaq9166]

Prof. The truth is that I will know nothing about analysis without a constant reminder of your Video tutorials. God bless your existence and your future generations. .

[@CrunchEconometrix]

Mr. Iziaq, I am encouraged by your feedback. Thanks so much, Sir 🙏❤️

[@asisomosfan666]

You're doing God's work here. You deserve the best. Thank you Professor.

[@CrunchEconometrix]

Hi Asiso, I'm encouraged by your words and prayers. Thanks!!!

[@Meera4980]

You are a blessing for students like us ❤

[@CrunchEconometrix]

Glad to hear this, Panchali! 💖

[@kabinehkpukumu6182]

God bless Dr. We can't appreciate enough for your great work

[@CrunchEconometrix]

You are welcome, Sir :)

[@mfigueira1303]

Thanky you very much for the class. I am writing my dissertation and am using GMM. Your videos are very helpful. Best.

[@CrunchEconometrix]

Thanks for the positive feedback, deeply appreciated! Wishing you the very best, Mauro!

[@EducationalCity2024]

how you measured the Wooldridge test of strict endogeneity

[@shrikantdethe9825]

Thank you so much, "collapse" option is so great, solved much of my problem!! 💌

[@CrunchEconometrix]

Glad to hear.

[@zuzia6021]

Hello! I did the GMM without collapse, i got Arellano-Bond test for AR(1) in first differences: z = -3.76 Pr > z = 0.000 Arellano-Bond test for AR(2) in first differences: z = -1.10 Pr > z = 0.273 ------------------------------------------------------------------------------ Sargan test of overid. restrictions: chi2(75) = 181.59 Prob > chi2 = 0.000 (Not robust, but not weakened by many instruments.) Hansen test of overid. restrictions: chi2(75) = 74.21 Prob > chi2 = 0.504 (Robust, but weakened by many instruments.) and with collapse i got the opposite. What does this mean and what should i do in this case? It says the instrument set is good in this case? And then i do it with collapse and i get : Arellano-Bond test for AR(1) in first differences: z = -2.82 Pr > z = 0.005 Arellano-Bond test for AR(2) in first differences: z = -1.13 Pr > z = 0.257 ------------------------------------------------------------------------------ Sargan test of overid. restrictions: chi2(9) = 52.19 Prob > chi2 = 0.000 (Not robust, but not weakened by many instruments.) Hansen test of overid. restrictions: chi2(9) = 17.85 Prob > chi2 = 0.037 (Robust, but weakened by many instruments.)

[@CrunchEconometrix]

No query here.

[@zuzia6021]

@@CrunchEconometrix Sorry, i dont understand? :O

[@zuzia6021]

@@CrunchEconometrix And thank you somuch for responding, your videos are a source of great relief and the only reason I manage to get through my studies. I am really grateful for what you have done :) I did not expect you to respond but was hopeful :D

[@CrunchEconometrix]

Hahahaha, no worries Upasana😊. My response rate is over 98%. Even if I don't have the answer I'll be honest with you. Gr8 feedback on my videos. Deeply appreciated! Please may I know from where (location) you are reaching me?

[@zuzia6021]

@@CrunchEconometrix

[@GarimaGupta]

thank you for covering this topic. I want to know what to do when AR(2) is coming significant? What can be possible manipulations?

[@CrunchEconometrix]

It implies the presence of 2nd order serial correlation. Kinda read more about this from online blogs, articles, articles, and textbooks.

[@lkadzo2880]

If possible, please could you cite the studdies that suggest only modest efficiency gains from two-step GMM?

[@CrunchEconometrix]

Kadzo, references are listed at the end of each GMM video.

[@adityarazpokhrel7626]

Thank you mam... clear concepts and hence it is very useful... Mam my request to you is, can you please make videos on GVAR, SVAR and FAVAR modelling with STATA or E views domain ?

[@CrunchEconometrix]

Thanks Aditya, for the suggested topics. I have noted them.

[@adityarazpokhrel7626]

Thank you mam, once again.

[@Lucyferandtheson003]

Hi Professor, thank you so much for making the whole GMM stuff so easy. Would you kindly clarify why we need to create year dummies and how we are using them in panel data please? Thank you

[@CrunchEconometrix]

Hi Oliver, kindly watch the video on "GMM Year Dummies" for clarification. Thanks.

[@Lucyferandtheson003]

@@CrunchEconometrix Thank you so much professor. I will make sure I direct all my students and friends to your channel.

[@Lucyferandtheson003]

@@CrunchEconometrix Hi prof, please assist me decipher this: I am getting this error when I try to run any of the system and difference GMM models >>> "mata: mata set matafavor speed, perm. / / / / / / invalid.r(198);" What could really be the problem? I checked the syntax and its correct

[@CrunchEconometrix]

Hi Oliver, kindly post this on Statalist.org for constructive feedback. The programmers will be able to assist you.

[@anggaraadhari1933]

thank you so much, from Indonesia

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@plzrelax9396]

HY Mam! Thank you so much for starting the CurnchEconometriscs for begginers. I saw most of your videos and these were amazing amazing. I understand the things clearly. Please make the video about viusalization of panel data. like how to plot the graph of the data. how we show the trend of a variable across income groups or regional graph

[@CrunchEconometrix]

Hi there, thanks for the positive feedback on my videos. Deeply appreciated. I also have videos on all the queries raised. Check my Panel data Playlist and you will see them all...the basics of panel data, graphs, scatterplots etc. So, keep watching, keep sharing...may I know your country or where you are reaching me from?

[@plzrelax9396]

Thank you Ma'am for sharing the information. I am M.Phil student from Pakistan

[@ilhanyarkan7140]

first of all, thank you so much for your perfect contributions on young researchers. That is an incomparable help for many researchers such as myself. And I want to ask you, ig we decide to run GMM estimator, do we need to employ a stationary test before? Is there any requirement to variables should be I(0) or I(1)?

[@ilhanyarkan7140]

and also, what happens if T is quite higher than N in a panel data? can we still run GMM?

[@CrunchEconometrix]

Hi Ilhan, kindly watch the introductory GMM video. Provides answers to your queries. Thanks

[@oluwaseunadeoyeoyebamiji3592]

I have so-far followed up on your lectures, thank you once again ma. I took your advice and I checked on my model, corrected and I attempted to run xtabond2, however, I probably made a mistake or perhaps STATA is proving otherwise difficult to use for me. My specification xtabond2 LogAgricGDPVALUEADDED l.LogAgricGDPVALUEADDED LogPrecipitation y*, gmm(LogNonAgricGDPvalueadded LogInvestmentGrosscapitalfo LogofSchooling Foreigndirectinvestmentnet MinningofGDP TradeopenessimportexportGD, lag (2,.) iv(LogPrecipitation)) nodiffsargan twostep robust orthogonal small but STATA says Lag limits must have two arguments. I can barely understand this. I was hoping you could like always be of help ma. Thanks

[@CrunchEconometrix]

Ur model is wrongly specified. Dislodge gmm(). Modify my dofile paying attention to endo, predetermined and exo variables. Advisable to watch the first video if you have not done so and jot some notes while at it.

[@HE-gw2gr]

Hello, I am watching from Turkey. I am grateful for your videos. I wanted to purchase your courses and do files from your website. However, the payment is not going through. Why is that? I have an urgent need for them.🥺🙏

[@CrunchEconometrix]

Hi there, I deactivated the link since 2019. Kindly know that due to abuse, unethical conduct, and several attempts to hack my Google drive, Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php The files don't cost much, just a token to maintain my website. Thanks for your patronage. Appreciated.

[@prashantgupta3123]

Could you please explain more about predetermined and endogenous variables?

[@CrunchEconometrix]

Hi Prashant, I already gave direct explanations. To save time, you may have to surf through textbooks and papers for details...but the truth is that the researcher makes the classification based on the study focus and theory. Thanks.

[@danielkrupah]

Hello Dr Adeleye, Please sorry to disturb you again. Please kindly find below this output and give your suggestions. One-Step Difference GMM. After trying different options since yesterday, I have been able to see two of independent variables significant. Please check the AR(2) and the Hansen test for me. Thank you Arellano-Bond test for AR(2) in first differences: z = 1.47 Pr > z = 0.143 ------------------------------------------------------------------------------ Sargan test of overid. restrictions: chi2(73) = 240.45 Prob > chi2 = 0.000 (Not robust, but not weakened by many instruments.) Hansen test of overid. restrictions: chi2(73) = 29.51 Prob > chi2 = 1.000 (Robust, but weakened by many instruments.) This

[@CrunchEconometrix]

Daniel, I'm constrained by time and my busy schedule to give personalized tutoring. Please follow my interpretations and adapt to your results. Thanks for your understanding.

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@shaxiyaali5576]

thank you very much for this video. I cannot find the do file for this video on your website. Could you please help me?

[@CrunchEconometrix]

Hi Shaxiyaali, due to abuse and unethical conduct Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php

[@emregokceli5087]

Hi, many thanks for the video. Please could you explain the internal and external instruments?

[@CrunchEconometrix]

Hi Emre, kindly watch the prerequisite video for the explanation. Thanks.

[@joylm9108]

Thank you professor for your videos. Can you use regional dummies instead of year dummies ?

[@CrunchEconometrix]

Hi Joy, thanks for the positive feedback, deeply appreciated 😊 Regional and year dummies capture different information. Kindly watch my videos on Error Component Models. Thanks

[@horveysenyosylvester4285]

Thank you very much, Prof. I have watched your videos many times and they are very helpful. Please I am running a panel data (N=63; T=4). AR(1) is not significant and there are no values for AR(2). Can I report only the AR(1) in my results? How do I correct the no values for AR(2)?

[@CrunchEconometrix]

Horvey, AR2 and Hansen J are the important tests. You can increase the years then re-estimate.

[@horveysenyosylvester4285]

Many thanks. I increased the years and it worked

[@mahinurmimi9782]

Dear Professor, Is it always necessary to use the lag value of the dependent variable as a regressor in GMM? Besides, in gmmstyle() can we use lag 2/3/4/5 of dependent variable?

[@CrunchEconometrix]

Hi Mahinur, kindly watch the introductory GMM video and read some of the literature on GMM (listed at the end of the GMM videos) for better understanding of the GMM technique. Thanks

[@simonetaddeo1935]

Dear Professor, could you give an advice to how interpret "Sargan test" p-value? A) If I have A Sargan test (p-value 0.110) and Hansen (p-value 0.125) B) If I have A Sargan test (p-value 0.000) and Hansen (p-value 0.125) C) If I have A Sargan test (p-value 0.050) and Hansen (p-alue 0.125) How could I interpret these result? Which one of the 3 options is best to accept for you?

[@CrunchEconometrix]

Hi Simone, from my videos the emphasis is on Hansen and not Sargan for obvious reasons which I explained in the introductory video. Both of them attest to the "goodness" of the instruments. Same interpretations to their respective p-values.

[@simonetaddeo1935]

@@CrunchEconometrix thank you for the response. So if I am in a situation where I have A Sargan test (p-value 0.000) and Hansen (p-value 0.125), which do you advice me to do?

[@CrunchEconometrix]

Simone, go with Hansen and interpret your results. It is more robust than the Sargan statistic.

[@LilyEaglandxxx98]

Hi there, thank you so much for these videos they have really helped out a lot! I have followed your advice, however in my model I get a sargan p-value of 0.000, what do you advise I do next? I follow the exact specification laid out by a highly regarded paper, with the same variables but different countries. However I only have 19 countries and 10-19 years of quarterly data (unbalanced). Thank you in advance for your response.

[@CrunchEconometrix]

Hi Lily, thanks for the positive feedback and remarks. Deeply appreciated! I'll suggest that you pay attention to your Hansen statistic as mentioned in the introductory video. Please may I know from where (location) you are reaching me?

[@LilyEaglandxxx98]

@@CrunchEconometrix thank you for this, I am from London! Also, I wanted to ask when using quarterly data, if it is still correct to generate yearly dummies? And when you refer to N > T, does T mean number of years or number of quarters (in the case when we have quarterly data)? Thank you.

[@CrunchEconometrix]

Hi Lily, in that case, you will generate quarterly dummies and T is the time dimension of your dataset.

[@NehaGupta-rn7sd]

Does instruments include all explanatory variables (main independent variables + controls)?

[@CrunchEconometrix]

Yes... though, not in all cases.

[@kefassidauruk7284]

Thanks for this clear and straight-forward video. I would like to ask how to decide which variable to put on internal instruments and external instruments? Is it based on our own assumption or there is an empirical way to do it?

[@CrunchEconometrix]

Thanks for the encouraging feedback. Deeply appreciated! The researcher decides either based on theory, similar studies, or intuition. Please may I know from where (location) you are reaching me?

[@kefassidauruk7284]

@@CrunchEconometrix Thanks for the answer. I'm from Indonesia

[@vitaltopics316]

Hello, thank you so much for the tutorials. My question is "how to select the instrument variables?"

[@CrunchEconometrix]

Hi King, you determine your instruments.

[@vitaltopics316]

@@CrunchEconometrix thank you so much for your prompt reply, but how to determine? is there a technique or criteria?

[@CrunchEconometrix]

Please read Roodman (2009, 2014) for details about instruments in GMM.

[@vitaltopics316]

@@CrunchEconometrix Noted! Thank you so so much

[@thicamvanha3567]

Thank you very much for sharing those fantastic videos, Prof. They are very clear and in a great order for the beginners to understand them well. May I ask a question? In applying both Different and System GMM on my dataset, the results generated the omitted coefficient on few years :). What could be the case and how to deal with that? And what should I do when using collapse option or changing/swapping the regressors and the Pvalue on Hansen test is still smaller than 0.2, and sometimes is (.).

[@CrunchEconometrix]

Hi Thi, thanks for the positive feedback on my videos. Deeply appreciated! 💕 But I don't know what you mean by "results generated omitted coefficients on few years". Can you be more specific?

[@sarahahmedchawsheen5455]

Hi, can I use GMM for a panel N

[@CrunchEconometrix]

Sarah, as explained in my GMM videos it is applicable N>T panels.

[@sarahahmedchawsheen5455]

@@CrunchEconometrix yes thanks.

[@AjuluOkeke]

How about using Eviews for the different categories of GMM?

[@CrunchEconometrix]

Hi Obiajulu, I find Stata more robust for GMM analysis than using EViews. Thanks

[@EducationalCity2024]

Great work Prof, how you measured the Wooldridge test of strict endogeneity,

[@CrunchEconometrix]

I have responded to you that I have no idea. You may want to check other online resources.

[@shirazshiraz8404]

Hi, should we generate year dummies? why should we use year dummies in dynamic panel model?

[@CrunchEconometrix]

Hi Shiraz, I answered your queries in my video on GMM year dummies. Kindly watch it, thanks.

[@justintusoe5661]

I am currently working on the impact of human capital development and economic growth in Sub Sahara Africa. My dependent variable is GDP per capita, whilst my independent variables are school enrolments( primary, secondary and tertiary), life expectancy, trade openness, FDI financial development, labour force, gross fixed capital formation, and the lagged value of GDP per capita. Using the SGMM, It is enough to lag only the dependent variable? Secondly, I have difficulty in differentiating between the empirical model from the estimation technique

[@CrunchEconometrix]

Hi Justin, I recall advising you on Facebook to watch the GMM series and your query shows that you have not done so. Kindly do as as I addressed some of these issues and you will find answers to others in the course of watching my videos on panel data analysis. Thanks.

[@drsaghirghauri5361]

In the lecture "How to Estimate One-Step Difference GMM" you said that Dofile is free and available at your Website, but unfortunately I could not find the free Dofile, kindly guide me, I need dofile in order to replicate my own model.

[@CrunchEconometrix]

Hi Dr. Ghauri, Due to abuse and unethical conduct Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php Thanks for the purchase, appreciated 🙏

[@LoanVuBich-q6c]

Thank you very much, Professor. I have a question. How can I find the do-file on your website?

[@CrunchEconometrix]

Hi there, here's the link cruncheconometrix.com/view/datashop.php

[@ramadhani9880]

​@@CrunchEconometrixwe can't access the web

[@CrunchEconometrix]

Hi Ramad, thanks for your enquiry. Kindly know that due to abuse and unethical conduct, Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php The files don't cost much, just a token to maintain my website. Thanks for your understanding and patronage.

[@EasyLearningMMA]

my coefficient of lagged value of the dependent variable is negative in gmm xtabond2 how make positive?

[@CrunchEconometrix]

Your result is a function of several factors: variables used, methodology and estimation technique.

[@parfaitberi3974]

Thank you very much, Professor. The papers you suggested in your previous video are mathematically demanding but this practical session actually demystifies the whole stuff for me. Just two quick question for you Professor, Should data in percentage form be logged? e.g. the Inflation rate, the growth rate of GDP. 2ndly, How many max number of variables do you recommend for a single model? Mercie, writing from South Africa but originally from Cameroon

[@CrunchEconometrix]

Thanks Beri for the encouraging feedback. Deeply appreciated! 💕 I'll say rate variables should not be logged but others may disagree as I've seen published papers in reputable journals use logs of inflation rates and growth rates. Perhaps, their models responded better with the logs of the variables. I'm always an apostle for parsimonious models. Not more than 4 variables. 1 depvar and 3 explanatory variables. My love to Cameroonians and South Africans alike. Kindly share my KZbin Channel link with your students, friends and academic community on social media for awareness. They'll learn some useful tips and skills too! 😊

[@HoaNguyen-zn9yb]

Thank you for the very useful video. Can I ask you a question, what y* here mean? Thank you so much!!

[@CrunchEconometrix]

Hi Hoa, thanks for the positive feedback. Deeply appreciated! The y* represents the year dummies. Thanks for watching and sharing, grateful!❤️

[@barhom11eriqat98]

Can I relay on the results without robust command? I have very good results when I ignore robust command, but very bad when I added robust, If not, what is your recommendations to solve this problem? Thanks

[@CrunchEconometrix]

Hi Barhom, this is quite dicey as the ROBUST option controls for heteroscedasticity. You can proceed with the one that gives you the best results provided no Reviewer queries your results. Thanks

[@barhom11eriqat98]

@@CrunchEconometrix Thank you professor 🙏🙏

[@entfe]

Hi Dr., thank you so much for your videos. They are very useful and helpful, and I wish you continue to thrive! I have a question about when you included inflation as an exogenous var.; what is this exactly instrumenting for (i.e. is there a certain variable it is instrumenting for)? thank you in advance

[@CrunchEconometrix]

None in particular. It is just included as part of the instruments.

[@yambahadurkarki2432]

Thank you.

[@CrunchEconometrix]

You're welcome, Yam!

[@enkii82]

Commands: 07:56 without collapse cmnd 09:40 include collapse cmd 10:30 include new var_inflation 11:38 exclude nodiffsargan and collapse cmd

[@CrunchEconometrix]

Hi Enk, may I know what your queries are? Indicating "time" alone are not helpful to your cause if you don't state what the issues are.

[@mohamedamiir6360]

Hi Prof! Could you please elaborate how you created year dummies in this example?

[@CrunchEconometrix]

Hi Mohd, I used Stata to generate them.

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@lalychugden6760]

Dear professor, could youu please share your do file? the website appears down. Thanks for your amaing help!

[@CrunchEconometrix]

Thanks for your enquiry. Kindly know that due to abuse and unethical conduct, Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php The files don't cost much, just a token to maintain my website. Thanks for your understanding and patronage.

[@superloula5106]

Hi Dr. , Hansen test of overid. restrictions: chi2(163) = 0.00 Prob > chi2 = 1.000 , please how can I correct this ?

[@CrunchEconometrix]

Hi Loula, there's no straight approach to correcting this. You may need to play around with the lags and instruments set.

[@superloula5106]

Thanks so much ،،💞❤️

[@ednono4540]

Hi Madam, Thank you very for this help. If I may ask, have you made the same estimation on Eviews please? if yes, I'll be grateful to have the link. Thank you for all videos. They are very useful

[@CrunchEconometrix]

Hi Emeline, thanks for the positive feedback on my videos. Deeply appreciated. But at the moment, I don't have GMM-EViews videos. Please may I know from where (location) you are reaching me?

[@ednono4540]

@@CrunchEconometrix Hi, Madam! I am in Belgium and a student at Louvain School of Management

[@CrunchEconometrix]

@@ednono4540 Awesome girl!❤️I'll appreciate it if you can share the link to my KZbin Channel with your friends and academic community in Belgium 🇧🇪. They will find the content helpful. Thanks 😊

[@barnonamondol5032]

Thank you so much for your tutorials. I have a question "How did you create 11 years dummies ?" In my analysis it has been shown that "xtabond2" as unrecognized command. why is that? can you please explain. Thanks in advance.

[@CrunchEconometrix]

Barnona, type "help xtabond2" in the Command Window and follow the Stata prompts to install the syntax. tab year, gen(yr)

[@barnonamondol5032]

@@CrunchEconometrix Thank you for your kind reply. it worked thanks to you. But after running one step difference year1 and year2 dummy variable have been dropped by Stata saying "Year1 dropped due to collinearity". will these affect my result? is there any solution for that? Thanks in advance.❤

[@CrunchEconometrix]

Not at all.

[@bangladeshisoulsinuk4993]

@@CrunchEconometrix what is an year dummy and wh yare we creating a year dummy

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@bangladeshisoulsinuk4993]

how did you get these instruments. I am trying to find out the effect of FDI on Gender inequality Index. I just used the lag of GII. What instruments shall I use, I am completely lost.

[@CrunchEconometrix]

I explained instrumentation in detail from the very first video of the GMM series. I advise you watch the clips in sequential order. Thanks

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@joffrimpong7720]

Please, how did you include your year dummy variables, l know by y*, but how did you arrive at it

[@CrunchEconometrix]

y* represents the year dummies. I showed this in most of my panel data videos. You may want to watch them.

[@joffrimpong7720]

@@CrunchEconometrix I would like to watch them and if please you can send me the link. Again you presented a dummy for all 11 years are you not falling into the dummy trap situation

[@CrunchEconometrix]

Kindly browse the Playlists for Panel Data videos. No dummy variable trap as a year dummy is omitted during estimation.

[@handecakr8775]

again it is a very useful video...

[@CrunchEconometrix]

Thanks for the encouraging feedback, Hande. Deeply appreciated! Please may I know from where (location) you are reaching me?

[@sumedhasinhamusic]

Madam, I am from India and had an interaction over email. Well, I tried to install xtabond2 using the techniques that you mentioned. But, it is still not running. Can you help? Well, I am doing a work on finance. The dependent var. is firm value and the independent variables include one-year lag of firm value and dividend policy as the key independent variables and others like leverage, liquidity, age as the control variables. The work also includes period dummies: 0 for the pre-financial crisis period and 1 for the post financial crisis dummies. In the DP model, how do I decide the pre-determined variables? For the instrumental variables, should we add one year lag for all the independent variables? Can you depict one case not using command but staring from statistics.....and then doing?

[@CrunchEconometrix]

Hi Sumedha, thanks for reaching out. 1) You have to install xtabond2 syntax using any of these three options: a) ssc install xtabond2 b) findit xtabond2 c) help xtabond2. I told you these via email. 2) Watch the preliminary video again to understand instruments in gmm( ) and those in iv( ). You make your assumptions on these classifications based on theory, logical reasoning and similar studies. 3) Yes, a lag of regressors can be used as instruments. Regards.

[@quegiang9185]

There is a small problem with the last part of the video: you removed the variable "infl" from IVstyle() but you did not remove it from declaration in the command

[@CrunchEconometrix]

Hi Vac, thanks for the observation. Put that is not a problem per se. GMM allows the exclusion of some regressors as IV instruments. You don't have to use all as instruments. Please may I know from where (location) you are reaching me?

[@quegiang9185]

@@CrunchEconometrix Hi Adeleye, I am from Vietnam. You made really good videos with detail instruction for beginners. If the exclusion of some regressors as IV instruments is not a big problem, do you think this is another way of reducing instruments and improve Hansen test? It would be great if further recommendation provided in your videos regarding the use of gmm (.., eq(l)) and gmm(.., eq(d)) as well as the application of xtabond2 when data has irregular spacing. Have a nice day

[@CrunchEconometrix]

@@quegiang9185 Thanks for the encouraging words. I'll keep doing my best for the global academic community. I tried to cover the basics of the GMM technique, you will have to read the references for more information. Yes, exclusion of regressors may improve the efficiency of the GMM estimator for some studies. I'll appreciate it if you can share the link to my KZbin Channel with your friends and academic community in Vietnam 🇻🇳. They will find the content helpful too 😊.

[@ElvisCalvinOwusu]

I am unable to access your website for the do files😒😒

[@CrunchEconometrix]

Hi Elvis, thanks for your enquiry. Kindly know that due to abuse and unethical conduct Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php The files don't cost much just a token to maintain my website. Thanks for your understanding and patronage.

[@asgharkamal4633]

HI, Prof, when N < T. what can do

[@CrunchEconometrix]

Asghar, kindly watch my panel ARDL videos.

[@asgharkamal4633]

@@CrunchEconometrix oky highly obliged

[@maryemn4840]

Great video! Thanks lots. Please I can not access the website. Where can I find the do-file? many thanks.

[@CrunchEconometrix]

Hi Maryem, thanks for the encouraging feedback. Deeply appreciated! Dofiles are no longer free but available on my website upon payment of a token after which you are allowed a one-time download. Here's the link cruncheconometrix.com.ng/shop/

[@maryemn4840]

@@CrunchEconometrix Ok! Thank you Professor!

[@saraherviani84]

hai, thank you for your interesting video, anyway can I download do file since i cant access your website?

[@CrunchEconometrix]

Hi Sarah, files are accessible on my website cruncheconometrix.com.ng/shop

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@tommasoiori6555]

Thank you so much for your videos! I found them extremely helpful. However, I have a doubt. When applying a Difference-GMM, can I use differenced lags of variables as instruments or do I need to use only lags in levels? My problem is that I only have a flow database, so I don’t have levels of variables, but just variations from year to year. Is it an error to use differences instead of levels as instruments? Thank you for your help.

[@CrunchEconometrix]

Hi Tommaso, thanks for the encouraging feedback on my videos. Deeply appreciated! I suggest you follow the steps shown for the difference GMM technique. Thanks.

[@shabnamnazir5]

When we use "noleveleq" constant term disappears from results?? Is it okay

[@CrunchEconometrix]

That's how the algorithm is programmed so, that shouldn't be a problem.

[@shabnamnazir5]

@@CrunchEconometrix But how we will interpret results of constant term then ??

[@CrunchEconometrix]

Give the usual interpretation attributed to the "intercept".

[@evaggeliasiopi4746]

thanks for all, when can I find a free version of stata13? I have already the version of stata15 and I can't find the way to follow your instructions. Thanks in advance

[@CrunchEconometrix]

Hi Eva, I have no idea about getting free v13. Why not use use the v15?

[@najibhamood1966]

Excuse me, how can I get the do file of these videos?

[@CrunchEconometrix]

Hi Najib, thanks for your enquiry. Kindly know that due to abuse and unethical conduct Stata dofiles used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php The files don't cost much just a token to maintain my website. Thanks for your understanding and patronage.

[@MultiSlicer123]

Hi, I would like to know if it is okay to use industry dummies as external instruments and how the syntax would be like? Thank you.

[@CrunchEconometrix]

You put the dummies in the iv( )

[@enkii82]

Why I didn't get the y_1 y_2 ... y_11 in the result? where is that in the command equation?

[@CrunchEconometrix]

Check the xtabond2 syntax. It's all there.

[@mustanggemini2156]

Hi Dr, if the p-value AR(2) is 0.9, is that acceptable? Or there is an optimal range of p-value for AR(2) and Hansen test. Thank you for your time to teach us👌

[@CrunchEconometrix]

Please watch the 1st video of the GMM series and note my explanations on the Hansen statistic.

[@zuzekandovela3872]

how to generate l.lngdp (lag of lag of gdp) in Stata, please help fam

[@CrunchEconometrix]

gen l.lngdp = L.Ingdp Or use l.lngdp directly into the model.

[@zuzekandovela3872]

@@CrunchEconometrix thank you Doc for your response and assistance in my research. Stata kept giving me “ l.lnhdi invalid name” I am using human development index (HDI) as a proxy for sustainable development in my research ( impact of Fintech in sustainable development). I ended up replacing the . With _ and I had gen l_lnhdi = L.lnhdi. I’m a bit confused because there is no values/data/observations on the do editor for this variable name😭. What Could I be doing wrong?. Thank you for all your assistance Crunch Fam

[@user_bongbong118]

Thank you for this nice tutorial video. May I ask a question? My model failed to reject Arellano-Bond test for AR(1). Its p-value is over 0.05. In that case, what should I do?

[@CrunchEconometrix]

Nothing to worry about, BK. Most importantly, AR(2) must NOT be statistically significant.

[@user_bongbong118]

Thank you! :)

[@2001aunti]

in the command (xtabond2......) there is (y*) what does it mean? it means dummy for year? I try to do GMM at the firm level with N=500 and T=12 but I do not know the command for xtabond2 completely. appreciate your response. Also, once I try to run Xtabond2, I need to set the variable, using xtset. Can you help me how to do this?

[@CrunchEconometrix]

Yes, y* represents the year dummies. If you devote some time to watch my 9 GMM videos, you will become familiar with the xtabond2 syntax. Dofiles are available upon payment, if you are interested. Thanks

[@jsonkaindume1391]

Hi Doc, How can I get the Do file?

[@CrunchEconometrix]

Due to unethical conduct and several attempts to hack my Google drive, some datasets and all Stata dofiles used for all my videos are no longer free but available on my website upon payment. Here's the link👇 cruncheconometrix.com/view/datashop.php Thanks.

[@EasyLearningMMA]

your do files are paid. how can i download them now?

[@CrunchEconometrix]

The download link will automatically appear on your dashboard. Please, when did you make the payment?

[@dontworry828]

countries: 6, years: 28 would there be a problem with that? and what is the meaning of year dummies?

[@CrunchEconometrix]

Please find the time to watch the 9 GMM videos in sequential order. They answer 98% of your queries. Thanks.

[@TYS22]

Hi, Pls, I am finding it hard to impose convex restriction ( like a1+a2=1) in GMM estimation on Stata. I will appreciate your response. Thanks

[@CrunchEconometrix]

No idea of what you intend doing.

[@1112-x9s]

Thank you for the video and the explanation . Please I have two question : the first one How many years panel data are required to apply system GMM? ( I have 10 countries and 8 years ) , the second question is about the years dummy variables , are they an obligation to introduce them in our panel

[@CrunchEconometrix]

Some are of the opinion that N must be sufficiently greater than T, but there is no uniform agreement to that. So, I'll say estimate your analysis and see what you have. Watch my video on GMM year dummies and you have the answer, to avoid repetition. Thanks.

[@1112-x9s]

@@CrunchEconometrix thank you a lot for your answer

[@CrunchEconometrix]

@@1112-x9s U're very welcome! 💕 Kindly share my Channel link with your students and academic networks! May I know where you are reaching me from?

[@1112-x9s]

@@CrunchEconometrix yes of course , I did it. Please professor , I didn't find the answer for my seconde question"about the years dummy variables , are they an obligation to introduce them in our panel" or when do we have to include them ??

[@CrunchEconometrix]

@@1112-x9s I explained the essence of year dummies in the video for that. Kindly watch it to know more. Thanks.

[@lemakargar6356]

Dear dr.Crunch, thanks for the usefull tutorials. i have a question. I am trying to run my model using one step GMM difference. I did a correlation test and I have the following results. what should I do now if some of the variables are correlated? . corr lngdppercapita lnmigrationinflow lnpopulation lncapitalstock lnhumancapital (obs=600) | lngdpp~a lnmigr~w lnpopu~n lncapi~k lnhuma~l -------------+--------------------------------------------- lngdpperca~a | 1.0000 lnmigratio~w | 0.4109 1.0000 lnpopulation | -0.1819 0.0782 1.0000 lncapitals~k | 0.1380 0.8263 0.2467 1.0000 lnhumancap~l | 0.4225 0.3561 -0.2014 0.1949 1.0000

[@CrunchEconometrix]

You can't put highly collinear variables in the same model. Drop one, or estimate a stepwise regression.

[@lemakargar6356]

@@CrunchEconometrix thank you

[@lemakargar6356]

Dear Dr. Crunch, if you don't mind me asking what do you mean by estimating a step-wise regression? it would be really helpful , because I can't drop the variable that's highly correlated as it's essential for my analysis. so I would like to do the later approach you mentioned, but don't know exactly what you mean by step-wise regression estimation. thank you very much.

[@CrunchEconometrix]

@@lemakargar6356 If A and B are highly correlated variables, you can get around this by estimating step-wise regressions. Example: first regression: model H, J, A; second regression: model H,J,B. That way, you're able to use both variables and the different impact of A and B is observed on the dependent variable.

[@CrunchEconometrix]

@@lemakargar6356 If A and B are highly correlated variables, you can get around this by estimating step-wise regressions. Example: first regression: model H, J, A; second regression: model H,J,B. That way, you're able to use both variables and the different impact of A and B is observed on the dependent variable.

[@dorstellgh6238]

Hello Dr, please is there a need to estimate unit root when using GMM

[@CrunchEconometrix]

Hi Nana, that's a tough one because there's time series GMM which I'm not familiar with (you may read up on that). But if you have less than 15 years, then testing for stationarity is not required.

[@yebouakouassi3694]

How to ınclude country and time dummies in GMM estimation ?

[@CrunchEconometrix]

You can only include time dummies which I showed in the GMM videos.

[@maryemn4840]

Hello. When I try to run the model, I get an error: "variable gmm not found' I am struggling with this error. Can someone please help? many thanks

[@CrunchEconometrix]

Do you have the xtabond2 syntax installed?

[@ogundipeayobolawole2971]

My Command is showing Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm. No observations. r(2000); what is the problem and solution ma

[@CrunchEconometrix]

Notify Stata Technical Team from the Help menu.

[@AGhulam73985]

Mam. Please help me . What is criteria for instruments ?

[@CrunchEconometrix]

Hi Aribah, I have responded to you on the other thread.

[@rochnaarora7478]

ma'am Do we need to run endogeneity tests before rrunning the GMM so as to decide among endogeneous and exogeneous regressors or we can just make an assumption about the exogeneity of certain regressors in our model? Hoping for a quick reply.

[@CrunchEconometrix]

Hi Rochna, I'll say GMM controls for endogeneity given the use of instruments during estimation.

[@rochnaarora7478]

@@CrunchEconometrix But ma'am when we run xtabond2 command that requires certain variables in the gmm list while others in the iv list so what will be the basis of putting variables in the iv list ( because as per my knowledge the exogeneous regressors are put under that category)

[@CrunchEconometrix]

I have responded to you on this on a different thread.

[@mariogonzalezsauri4632]

Hi there, Is it correct to have a first differenced panel for the two models (first differenced and system) ? Or is the input a regular panel and xtabond2 transforms it into a first differenced model.

[@CrunchEconometrix]

Hi Mario, just follow the procedure as shown in the video. The xtabond does all that at the background.

[@mariogonzalezsauri4632]

@@CrunchEconometrix gracias you are amazing. Hugs from NL.

[@CrunchEconometrix]

Thanks Mario!🙏

[@hllp2497]

Do we always need to generate year dummy in GMM regression.

[@CrunchEconometrix]

It is advisable to. it controls for the variation of the depvar in the data. Watch my video on that to find out more.

[@hllp2497]

Bosede Ngozi Adeleye thank you

[@SL-vd2dj]

Are the Do-files still free on your website? I can't seem to find the free version.

[@CrunchEconometrix]

No SL, dofiles are no longer free but available on my website upon payment of a token fee. Here's the link to the STORE cruncheconometrix.com.ng/shop/

[@lemakargar6356]

dear Dr.Crunch, is creating year dummies essential? and why?

[@CrunchEconometrix]

It controls for time variations. You can always get more online information on the essence of year dummies.

[@beambeam272]

@@CrunchEconometrix Dr. Crunch how to create year dummies essential?

[@CrunchEconometrix]

@@beambeam272 Are you making a comment or asking a question?

[@beambeam272]

​@@CrunchEconometrix I am asking a question because I don't understand how to you create year dummies essential

[@CrunchEconometrix]

@@beambeam272 Code is in the dofile.

[@kotikkompot]

Hello, I have some problems with gmm, when I run it I have no values for AR(2) and Hansen test, moreover, chi sq degrees of freedom is -2 for Sargan and Hansen tests. Can someone help me please

[@CrunchEconometrix]

Hi Michael, I advise that you watch the entire 9 GMM series for proper understanding of the technique. From the different simulations shown, you will know how to correct these issues. Thanks.

[@kotikkompot]

Thanks, I did some manipulations and figured out that the problem was with dummies, however, i still do not understand why i should not use all of them (like i have 4 time periods, so i should use just 3 or 2 dummies?). Why is that so?

[@CrunchEconometrix]

Due to "dummy variable trap" you cannot use all the dummy variables. You have to drop one...and the dropped dummy represents the intercept (constant) of the model.

[@AGhulam73985]

I don't understand equation What are instruments ??? How do u we select them ?

[@CrunchEconometrix]

Hi Aribah, relax make sure you watch the video on "Understanding GMM" if you haven't done so. GMM is a bit complex procedure so you have to take the learning process in steps and read journal articles that used the procedure. You make the assumptions concerning the categorisation of your variables, mostly from the theory you are using. After which you have defined the endo and predet variables, the rest will be exo variables which are put in the iv( ) while the endo and predet variables are included in the gmm( ).

[@AGhulam73985]

Thank you very much

[@mustanggemini2156]

Dr , when I add y* (year dummy) for diff and system gmm, one of my indep var coefficient is omitted. May I know why? Thanks in advance

[@CrunchEconometrix]

Most times Stata drops a year variable to avoid the "dummy variable trap". Check the top of the output to see why the explvar is dropped.

[@mustanggemini2156]

CrunchEconometrix Despite adding y* ( year dummies) or i.year for diff and system GMM, indep var coefficient is still omitted. No indication stated on the Output of regressions. AR(1), AR(2) and Hansen are good based on respective null. I wish I could capture the output for your review if that possible. Without year dummy, my models are good.

[@CrunchEconometrix]

Alright. If you preferred the model w/o year dummies proceed but note it in your paper for the readership to know why year dummies are omitted.

[@mustanggemini2156]

CrunchEconometrix noted Dr. I guess that’s the right thing to do as you mentioned. Again, I really appreciate your input, and thank you so much for being so helpful. Have a good day Dr👌

[@ogundipeayobolawole2971]

HELLO Prof. My command is giving out no mata

[@CrunchEconometrix]

Hi Ayo, I honestly can't tell why that is. Please send a mail to Stata Technical Team. Use the Help menu.

[@juabinmatey8503]

VIDEOS ARE NOT DOWNLOADABLE, WHY?

[@CrunchEconometrix]

Juabin, I'd prefer you watch the clips online to improve my viewing times and income earned instead of downloading them.

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@CrunchEconometrix]

Bismillah, I have responded to you on this on a different thread.

[@sidrakhan5608]

How can I download do file please help me

[@CrunchEconometrix]

Hi Sidra, Stata dofiles as mentioned in the video are available on my website.

[@shiyixu3249]

@@CrunchEconometrix hi， I didn't find dofile on the website, May i ask where can I find it ^-^

[@CrunchEconometrix]

Hi Shiyi, kindly click on cruncheconometrix.com/resources

[@shabnamnazir5]

if anybody have do files for free please share them with me I have not much money to purchase the files

[@CrunchEconometrix]

Dofiles are available on my website upon payment. Here's the link cruncheconometrix.com.ng/shop

[@fusonyBD]

It's a total lie that the do file is available free of cost.

[@CrunchEconometrix]

FusonyBD, I will advise you calm down and watch your choice of words. I made datasets and dofiles free-of-charge when I began the Channel in January 2018. I removed the privilege in August 2019 due to unethical conduct and several attempts to hack my Google drive. So, if any serious researcher needs resource materials, such will have to pay a token. Some datasets are free but all Stata dofiles are available on purchase. If you are interested here's the link cruncheconometrix.com/view/datashop.php

[@fusonyBD]

@@CrunchEconometrix Then, you have to edit your video. In the video, you have said that the do file is available free of charge. That is why, I have said that it's a lie. When people watch, you make money for that. In addition, you are compelling people to pay you extra. I wouldn't have reacted, had I not seen a statement that you are making is not correct and you are doing opposite.

[@CrunchEconometrix]

FusonyBD, I really do not owe you any explanation beyond what I said. Besides, it is NOT compulsory you use my datasets or Stata dofiles. If you are keen on learning, kindly avail yourself of codes shown on the screen vis-a-vis using your datasets. Thanks.

[@jahnavipratishtha]

Hello, I've been watching your videos since my undergrad! You are truly a lifesaver. I am trying to run a first difference GMM regression and i am getting an error message from stata. My code is: xtabond2 lFLFP l.lFLFP lGDPpc lGDPpc2 lFertilityRt lUnempl lLiteracyRt lLifeExpectancy i.Year, gmm(l.lFLFP, collapse) /// iv(lFertilityRt lUnempl lLiteracyRt lLifeExpectancy i.Year)/// noleveleq nodiffsargan robust small i.Year is my year dummy variable. Could you please help??

[@CrunchEconometrix]

You may want to adapt the GMM code I used for your analysis. The code is visibly shown in the clips.