FamaMacBeth regression
Moderators: EViews Gareth, EViews Moderator, EViews Esther

 Posts: 5
 Joined: Tue Mar 07, 2017 2:59 pm
Re: FamaMacBeth regression
Hi
I have read all of the comments on this thread and found a lot of them useful, so thanks.
I have an issue with the FM Model though. In the example files provided by the AddIn, the FMB regression is run using SMB, HML and Mkt_R as independent variables (IVs). Some other studies and examples in this thread also do the same. They therefore seek to explain how a portfolio is exposed to these various riskfactors (usually provided by K. French's website).
However, some other studies seek to assess the monotonic, or idiosyncratic, variation in stock returns. They tend to use market capitalisation (size), booktomarket ratio (value), momentum and market beta as the independent variables. The riskfactors (IVs) thus all include data for each firm. Earlier in this thread, it was mentioned that the riskfactors cannot currently take into account multiple series.
For some context, I am seeking see whether the variation in stock returns can be explained by the environmental performance of a firm  I understand that the FamaFrench and FamaMacbeth theories are the best ways to utilise control variables and make a robust analysis.
My questions are therefore the following:
1. Is the FamaMacbeth package on EViews solely for the use of K. French riskfactors? i.e. HMB, SML etc.
2. Am I overthinking everything...perhaps somebody knows an easy solution for my research question.
3. The FamaMacbeth method, as far as I am aware, is useful for assessing monotonic relationships. However, this relies on supplying independent variables that are firm specific (matrixform). It seems to me that the AddIn performs the FamaMacbeth procedure but does not allow for the analysis of individual stocks. Looking at others research, I am therefore not replicating their studies because my IVs are the general K. French factors and not the firmspecific factors..?
I greatly appreciate any support! For reference, my reference papers are mainly Manescu (2011) and Galema et al (2008).
I have read all of the comments on this thread and found a lot of them useful, so thanks.
I have an issue with the FM Model though. In the example files provided by the AddIn, the FMB regression is run using SMB, HML and Mkt_R as independent variables (IVs). Some other studies and examples in this thread also do the same. They therefore seek to explain how a portfolio is exposed to these various riskfactors (usually provided by K. French's website).
However, some other studies seek to assess the monotonic, or idiosyncratic, variation in stock returns. They tend to use market capitalisation (size), booktomarket ratio (value), momentum and market beta as the independent variables. The riskfactors (IVs) thus all include data for each firm. Earlier in this thread, it was mentioned that the riskfactors cannot currently take into account multiple series.
For some context, I am seeking see whether the variation in stock returns can be explained by the environmental performance of a firm  I understand that the FamaFrench and FamaMacbeth theories are the best ways to utilise control variables and make a robust analysis.
My questions are therefore the following:
1. Is the FamaMacbeth package on EViews solely for the use of K. French riskfactors? i.e. HMB, SML etc.
2. Am I overthinking everything...perhaps somebody knows an easy solution for my research question.
3. The FamaMacbeth method, as far as I am aware, is useful for assessing monotonic relationships. However, this relies on supplying independent variables that are firm specific (matrixform). It seems to me that the AddIn performs the FamaMacbeth procedure but does not allow for the analysis of individual stocks. Looking at others research, I am therefore not replicating their studies because my IVs are the general K. French factors and not the firmspecific factors..?
I greatly appreciate any support! For reference, my reference papers are mainly Manescu (2011) and Galema et al (2008).

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
What exactly are you asking?
The answer to your question #1 is No.
The answer to your question #1 is No.

 Posts: 5
 Joined: Tue Mar 07, 2017 2:59 pm
Re: FamaMacBeth regression
I am asking if the riskfactors can be a matrix?
I have added my own stock returns into the fm_example workspace. My files are called er_01, er_02 etc. I can then perform the FMB regression: fmb er* hml smb mkt_rf. Because the risk factors (hml, smb, mkt_rf) are fixed for all dependent variables, er* is therefore regressed against fixed risk factors.
Is it possible to change the risk factors so that they are in matrix form? So that every er_# is regressed against its corresponding value, size and market beta?
I have attached the explanation in the academic paper that explains the use of firmspecific independent factors, and also proof of my FMB regression output with the K.French factors. This might help explain my confusion.
I have added my own stock returns into the fm_example workspace. My files are called er_01, er_02 etc. I can then perform the FMB regression: fmb er* hml smb mkt_rf. Because the risk factors (hml, smb, mkt_rf) are fixed for all dependent variables, er* is therefore regressed against fixed risk factors.
Is it possible to change the risk factors so that they are in matrix form? So that every er_# is regressed against its corresponding value, size and market beta?
I have attached the explanation in the academic paper that explains the use of firmspecific independent factors, and also proof of my FMB regression output with the K.French factors. This might help explain my confusion.
 Attachments

 FMB Regression.png (128.63 KiB) Viewed 1544 times

 Manescu (2011).png (161.75 KiB) Viewed 1544 times

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
Have you read the addin documentation?

 Posts: 5
 Joined: Tue Mar 07, 2017 2:59 pm
Re: FamaMacBeth regression
I have read all of it yes. I have read Fama & Macbeth (1973) and Fama & French (1992; 1993) too. Perhaps I am missing something obvious  I have not been taught this stuff formally, only selftaught.

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
If you are asking whether you can force the addin to perform FamaMacBeth on multiple sets of risk factors at the same time, the answer is no. FamaMacBeth examines the risk premia from exposure to a common set of factors, not different factors for each return series. I suspect there is something you are not understanding about your problem.

 Posts: 5
 Joined: Tue Mar 07, 2017 2:59 pm
Re: FamaMacBeth regression
Thank you for the swift replies, I really appreciate the support.
Okay, I now understand what you are saying.
In Fama & French (1992) "The CrossSection of Expected Stock Returns" they seem to use multiple sets of risk factors for each return series. Is this correct?
Are you saying the addin can only perform and replicate the Fama & French (1993) paper "Common Risk Factors...", using common risk factors (SMB, HML)?
Therefore, I cannot use firmspecific risk factors in this addin. Correct?
Okay, I now understand what you are saying.
In Fama & French (1992) "The CrossSection of Expected Stock Returns" they seem to use multiple sets of risk factors for each return series. Is this correct?
Are you saying the addin can only perform and replicate the Fama & French (1993) paper "Common Risk Factors...", using common risk factors (SMB, HML)?
Therefore, I cannot use firmspecific risk factors in this addin. Correct?

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
You should read both papers, particularly FF92, more thoroughly.
The addin works the way it is described in the documentation.
The addin works the way it is described in the documentation.

 Posts: 5
 Joined: Tue Mar 07, 2017 2:59 pm
Re: FamaMacBeth regression
Hi Again
Could you explain how the code sets the number of lags for the Newey and West (1987) adjustment?
Is it automatically 4(T/100)^a (apparently what many econometrics software use)?
T = number of periods, a = 2/9 when using Bartlett kernel.
Thank you.
Could you explain how the code sets the number of lags for the Newey and West (1987) adjustment?
Is it automatically 4(T/100)^a (apparently what many econometrics software use)?
T = number of periods, a = 2/9 when using Bartlett kernel.
Thank you.

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
You can find more information about EViews standard errors at http://www.eviews.com/help/helpintro.ht ... rrors.html and for the addin in the output.

 Posts: 2
 Joined: Wed Apr 12, 2017 2:30 am
Re: FamaMacBeth regression
Hi,
i have to run several famamacbeth regressions for my factors.
I'm not that familiar with eviews but i think i understand how your code works.
i wanted to test if the hml factor yields the same risk premium and tstat as it did in your example. I did it for a 30 industry and 49 industry portfolio and my results don't make any sense.... (slightly negative and insignificant, while hml in your example is positive and has a tstat >2)
I think the issue is my portfolio... Did i write it in the right way? series* is my 49industry portfolio and pr* my 30 industry portfolio.
Thanks for your help.
i have to run several famamacbeth regressions for my factors.
I'm not that familiar with eviews but i think i understand how your code works.
i wanted to test if the hml factor yields the same risk premium and tstat as it did in your example. I did it for a 30 industry and 49 industry portfolio and my results don't make any sense.... (slightly negative and insignificant, while hml in your example is positive and has a tstat >2)
I think the issue is my portfolio... Did i write it in the right way? series* is my 49industry portfolio and pr* my 30 industry portfolio.
Thanks for your help.
 Attachments

 regression_inputdata_20170412.wf1
 8 factors + ff5 + 2 industry portfolios
 (660.03 KiB) Downloaded 29 times

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
I'm not sure what you mean. The example uses rmkt, not hml.

 Posts: 2
 Joined: Wed Apr 12, 2017 2:30 am
Re: FamaMacBeth regression
Okay, my question is not regarding the factors, but rather about the portfolio creation. In your file you name them 11 to 15 and than 21to 25, probably because it is a 5x5 portfolio. my question is if the way you name the portfolios has an impact on the fama macbeth regression.

 EViews Developer
 Posts: 89
 Joined: Thu Apr 18, 2013 8:37 am
Re: FamaMacBeth regression
It's not clear why you think the portfolio names would impact the regression. The portfolio names in the example are arbitrary (I downloaded the series from Ken French's data library and simply renamed them  the frenchdata addin is an easy way to download). Furthermore the same set of factors is used for each regression in the initial FamaMacbeth step, for example.
Re: FamaMacBeth regression
Hi everyone
I'm dealing with fama mcbeth regression but i cant realize the procedure.
In fm regression i want to regress monthly returns on operating profitability. I have 100 stocks with monthly return data and annual operating profitability data for ten years. and i also have 3 control variables (b/m , r1,1 , r12,2). So please help me how to fit these data on fm regression because returns data are monthly but operating profitability data are annual. I dont know how can i relate monthly data with annual...
having your assistance would be of great help. thanks.
I'm dealing with fama mcbeth regression but i cant realize the procedure.
In fm regression i want to regress monthly returns on operating profitability. I have 100 stocks with monthly return data and annual operating profitability data for ten years. and i also have 3 control variables (b/m , r1,1 , r12,2). So please help me how to fit these data on fm regression because returns data are monthly but operating profitability data are annual. I dont know how can i relate monthly data with annual...
having your assistance would be of great help. thanks.
Who is online
Users browsing this forum: No registered users and 5 guests