Re: [R] Hausman test

[David Winsemius]

> On Feb 11, 2018, at 8:29 AM, PAOLO PILI  wrote:
> 
> you are right about the 3rd line but it doesn't help me for my problem. I
> remove the 3rd line but there is still the same problem:
> 
> Error in solve.default (dvcov):
>   the system is numerically unique: reciprocity condition value =
> 1.63418e-19

That suggests inclusion of too many categorical (factor) variables relative to 
the sample size in the predictor variables. Use tabular methods to investigate. 
Unable to be more specific in the absence of a proper description of the data 
situation.

-- 
David.
> 
> Paolo
> 
> 2018-02-11 16:54 GMT+01:00 Bert Gunter :
> 
>> Note the typo in your 3rd line: data <
>> 
>> Don't  know if this means anything...
>> 
>> Bert
>> 
>> 
>> 
>> On Feb 11, 2018 7:33 AM, "PAOLO PILI"  wrote:
>> 
>>> Hello,
>>> 
>>> I have a problem with Hausman test. I am performing my analysis with these
>>> commands:
>>> 
 library(plm)
 data<-read.csv2("paolo.csv",header=TRUE)
 data<
>>> pdata.frame(data,index=c("FIRM","YEAR"),drop.index=TRUE,row.names=TRUE)
 
>>> RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+NGROWTH+TU
>>> RN+GPROF+GPROF2
 
>>> grun.fe<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROW
>>> TH+NGROWTH+TURN+GPROF+GPROF2,data=data,model="within")
 grun.re
>>> <-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+NGRO
>>> WTH+TURN+GPROF+GPROF2,data=data,model="random")
 
>>> gw<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+
>>> NGROWTH+TURN+GPROF+GPROF2,data=data,model="within")
 
>>> gr<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+
>>> NGROWTH+TURN+GPROF+GPROF2,data=data,model="random")
 phtest(gw,gr)
>>> 
>>> I got this answer:
>>> 
>>> Error in solve.default(dvcov) :
>>> 
>>> how can I solve this problem?
>>> 
>>> Thank you
>>> 
>>>[[alternative HTML version deleted]]
>>> 
>>> __
>>> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>> PLEASE do read the posting guide http://www.R-project.org/posti
>>> ng-guide.html
>>> and provide commented, minimal, self-contained, reproducible code.
>>> 
>> 
> 
>   [[alternative HTML version deleted]]
> 
> __
> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

David Winsemius
Alameda, CA, USA

'Any technology distinguishable from magic is insufficiently advanced.'   
-Gehm's Corollary to Clarke's Third Law

__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test

[PAOLO PILI]

you are right about the 3rd line but it doesn't help me for my problem. I
remove the 3rd line but there is still the same problem:

Error in solve.default (dvcov):
   the system is numerically unique: reciprocity condition value =
1.63418e-19

Paolo

2018-02-11 16:54 GMT+01:00 Bert Gunter :

> Note the typo in your 3rd line: data <
>
> Don't  know if this means anything...
>
> Bert
>
>
>
> On Feb 11, 2018 7:33 AM, "PAOLO PILI"  wrote:
>
>> Hello,
>>
>> I have a problem with Hausman test. I am performing my analysis with these
>> commands:
>>
>> > library(plm)
>> > data<-read.csv2("paolo.csv",header=TRUE)
>> > data<
>> pdata.frame(data,index=c("FIRM","YEAR"),drop.index=TRUE,row.names=TRUE)
>> >
>> RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+NGROWTH+TU
>> RN+GPROF+GPROF2
>> >
>> grun.fe<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROW
>> TH+NGROWTH+TURN+GPROF+GPROF2,data=data,model="within")
>> > grun.re
>> <-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+NGRO
>> WTH+TURN+GPROF+GPROF2,data=data,model="random")
>> >
>> gw<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+
>> NGROWTH+TURN+GPROF+GPROF2,data=data,model="within")
>> >
>> gr<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+
>> NGROWTH+TURN+GPROF+GPROF2,data=data,model="random")
>> > phtest(gw,gr)
>>
>> I got this answer:
>>
>> Error in solve.default(dvcov) :
>>
>> how can I solve this problem?
>>
>> Thank you
>>
>> [[alternative HTML version deleted]]
>>
>> __
>> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posti
>> ng-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test

[Bert Gunter]

Note the typo in your 3rd line: data <

Don't  know if this means anything...

Bert



On Feb 11, 2018 7:33 AM, "PAOLO PILI"  wrote:

> Hello,
>
> I have a problem with Hausman test. I am performing my analysis with these
> commands:
>
> > library(plm)
> > data<-read.csv2("paolo.csv",header=TRUE)
> > data<
> pdata.frame(data,index=c("FIRM","YEAR"),drop.index=TRUE,row.names=TRUE)
> >
> RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+NGROWTH+
> TURN+GPROF+GPROF2
> >
> grun.fe<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+
> PGROWTH+NGROWTH+TURN+GPROF+GPROF2,data=data,model="within")
> > grun.re
> <-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+PGROWTH+
> NGROWTH+TURN+GPROF+GPROF2,data=data,model="random")
> >
> gw<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+
> PGROWTH+NGROWTH+TURN+GPROF+GPROF2,data=data,model="within")
> >
> gr<-plm(RECEIV~LSIZE+LAGE+LAGE2+CFLOW+STLEV+FCOST+
> PGROWTH+NGROWTH+TURN+GPROF+GPROF2,data=data,model="random")
> > phtest(gw,gr)
>
> I got this answer:
>
> Error in solve.default(dvcov) :
>
> how can I solve this problem?
>
> Thank you
>
> [[alternative HTML version deleted]]
>
> __
> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/
> posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Achim Zeileis]

On Mon, 12 Sep 2016, Ding, Jie Ding (NIH/NIA/ERP) [F] wrote:


Dear Achim,

Sorry to have disturbed you. I have encountered a problem  when computing 
Hausman test statistics (i.e. p values)  in R to compare OLS and 2SLS models.

The problem is a discrepancy between the two p-value outputs from the "manual approach (by hand)" 
and the " diagnostics argument" in the "AER" library, respectively.

With respect to manual approach, I used the following codes:

cf_diff<-coef(ivreg)-coef(olsreg)
vc_diff<-vcov(ivreg)-vcov(olsreg)
x2_diff<-as.vector(t(cf_diff)%*% solve(vc_diff)%*%cf_diff)
pchisq(x2_diff,df=2,lower.tail=FALSE)


For diagnostic approach, I applied the following:

summary(ivreg, vcov = sandwich, df = Inf, diagnostics = TRUE)


However, p-value from the manual approach is always much larger than the 
diagnostic approach, e.g.  0.329 vs. 0.138


I would expect the values should be the same. Your advice would be 
highly appreciated.


The Wu-Hausman test in ivreg() follows the auugmented regression approach 
that is also used by Stata. This regresses the endogenous variable on the 
instruments and includes the fitted values in an OLS regression. The test 
is then a simple Wald test, see:

http://www.stata.com/support/faqs/statistics/durbin-wu-hausman-test/


With very best wishes,
Jennifer



[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Ding, Jie Ding (NIH/NIA/ERP) [F]]

Dear Achim,

Sorry to have disturbed you. I have encountered a problem  when computing 
Hausman test statistics (i.e. p values)  in R to compare OLS and 2SLS models.

The problem is a discrepancy between the two p-value outputs from the "manual 
approach (by hand)" and the " diagnostics argument" in the "AER" library, 
respectively.

With respect to manual approach, I used the following codes:

cf_diff<-coef(ivreg)-coef(olsreg)
vc_diff<-vcov(ivreg)-vcov(olsreg)
x2_diff<-as.vector(t(cf_diff)%*% solve(vc_diff)%*%cf_diff)
pchisq(x2_diff,df=2,lower.tail=FALSE)


For diagnostic approach, I applied the following:

summary(ivreg, vcov = sandwich, df = Inf, diagnostics = TRUE)


However, p-value from the manual approach is always much larger than the 
diagnostic approach, e.g.  0.329 vs. 0.138

I would expect the values should be the same. Your advice would be highly 
appreciated.

With very best wishes,
Jennifer



[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test trouble - plm

[TDix]

Might have just solved my own problem team!

I assumed that the issue here was the replicated samples, and so added a
column and gave a number to each replicate.

R seemed to like this and was happy to run the test!

A significant result tells me that the fixed effects model is the most
preferable model to explain the variation seen in my data.

Unless I am doing/assuming something wrong here that you can see then I
might well have solved my own problem.

Let me know if you have any thoughts :)

Cheers



--
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-Test-trouble-plm-tp4709990p4709992.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test in R

[Millo Giovanni]

Hello.
Well said Joshua. May I add that in an OLS context (which i take as
also meaning: no panel structure) what you probably want to do is the
so-called Durbin-Wu-Hausman test for endogeneity, as explained e.g.
here:

http://kurt.schmidheiny.name/teaching/iv2up.pdf

see Section 11 for the idea, and 13 for the R implementation.

Best wishes,
Giovanni

-- original message 
Date: Sun, 28 Oct 2012 16:03:43 -0700
From: Joshua Wiley jwiley.ps...@gmail.com
To: fxen3k f.seha...@gmail.com
Cc: r-help@r-project.org
Subject: Re: [R] Hausman test in R
Message-ID:

canz9z_+2k3qwazrazqz09nsfaj_431a2ylrpgswvnbo6pon...@mail.gmail.com
Content-Type: text/plain

Hi,

I can think of no reason a Hausman test could not be used for OLS---it
is a
comparison of vectors of coefficients from different models usually
assumed
to produce similar estimates under certain conditions.  Dissimilarity is
taken as indicative of a lack of some or all the conditions required for
the two models to yield similar parameters.
I suggest you look at the plm and systemfit packages.  They have many
functions for OLS, 2SLS, tests of endogeneity, etc.  The plm (and maybe
systemfit?) package also has a vignette which is a good thing to read.
It
has a lot of useful information on the code and examples of comparing
different types of models, that you may find instructive.

Hope this helps,

Josh


On Sun, Oct 28, 2012 at 1:33 PM, fxen3k f.seha...@gmail.com wrote:

 Hi there,

 I am really new to statistics in R and statistics itself as well.
 My situation: I ran a lot of OLS regressions with different
independent
 variables. (using the lm() function).
 After having done that, I know there is endogeneity due to omitted
 variables. (or perhaps due to any other reasons).
 And here comes the Hausman test. I know this test is used to identify
 endogeneity.
 But what I am not sure about is: Can I use the Hausman test in a
simple
 OLS
 regression or is it only possible in a 2SLS regression model? And if
it
 is
 possible to use it, how can I do it?

 Info about the data:

 data = lots of data :)

 x1 - data$x1
 x2 - data$x2
 x3 - data$x3
 x4 - data$x4
 y1 - data$y1

 reg1 - summary(lm(y1 ~ x1 + x2 + x3 + x4))

 Thanks in advance for any support!



 --
 View this message in context:
 http://r.789695.n4.nabble.com/Hausman-test-in-R-tp4647716.html
 Sent from the R help mailing list archive at Nabble.com.

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.




-- 
Joshua Wiley
Ph.D. Student, Health Psychology
Programmer Analyst II, Statistical Consulting Group
University of California, Los Angeles
https://joshuawiley.com/

--- end original message -

 
Ai sensi del D.Lgs. 196/2003 si precisa che le informazi...{{dropped:12}}

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test in R

[fxen3k]

Given my acknowledged statistical ignorance, I tried to find a *solution
*in this forum...
And this is not primarily a statistical issue, it is an issue about the
Hausman test in the R environment. 

I cannot imagine, no one in this forum has ever done a Hausman test on OLS
regressions.
I read in the systemfit package and found only this example referring to
2SLS and 3SLS regressions: 

data( Kmenta )
eqDemand - consump ~ price + income
eqSupply - consump ~ price + farmPrice + trend
inst - ~ income + farmPrice + trend
system - list( demand = eqDemand, supply = eqSupply )
## perform the estimations
fit2sls - systemfit( system, 2SLS, inst = inst, data = Kmenta )
fit3sls - systemfit( system, 3SLS, inst = inst, data = Kmenta )
## perform the Hausman test
h - hausman.systemfit( fit2sls, fit3sls )
print( h )




--
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-test-in-R-tp4647716p4647774.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test in R

[John C Frain]

On 29 October 2012 16:56, fxen3k f.seha...@gmail.com wrote:

snip

If we are talking about the same test a Hausman test can not be
applied to OLS regressions.  As you have already been told you must
have two estimates of the same set of coefficients to do a Hausman
test.

Suppose that you do OLS  and an IV estimates of a particular
regression you will get twu estimates of the coefficients in the
model. If the disturbances are not correlated with the explanatory
variables (no endogeneity) the two sets of coefficients will no be
similar.  If there is endogeneity the coefficients will be different.
The Hausman test is a test of the null that the coefficients are not
different.   If the null is accepted you will probably accept the OLS
regression. If the null is rejected you may consider the IV estimate.

A Hausman test is applicable in many other situations (fixed v random
effects etc.)  You may have problems with the estimate of the
covariance matrix used in the test as on occasion as, due to numerical
problems, the estimates of that matrix are not always positive
definite.

Most intermediate level econometrics textbooks will have a good
account of the Hausman test. Green(2012), Econometric Analysis 7th
edition, Prentice Hall. contains a comprehensive discussion of these
matters which you might read.  It is not easy but if you master the
basic concepts there, your questions about their implementation in R
are likely to be answered on this forum.

Best Regards

John

 I cannot imagine, no one in this forum has ever done a Hausman test on OLS
 regressions.
 I read in the systemfit package and found only this example referring to
 2SLS and 3SLS regressions:

 data( Kmenta )
 eqDemand - consump ~ price + income
 eqSupply - consump ~ price + farmPrice + trend
 inst - ~ income + farmPrice + trend
 system - list( demand = eqDemand, supply = eqSupply )
 ## perform the estimations
 fit2sls - systemfit( system, 2SLS, inst = inst, data = Kmenta )
 fit3sls - systemfit( system, 3SLS, inst = inst, data = Kmenta )
 ## perform the Hausman test
 h - hausman.systemfit( fit2sls, fit3sls )
 print( h )




 --
 View this message in context: 
 http://r.789695.n4.nabble.com/Hausman-test-in-R-tp4647716p4647774.html
 Sent from the R help mailing list archive at Nabble.com.

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.



-- 
John C Frain
Economics Department
Trinity College Dublin
Dublin 2
Ireland
www.tcd.ie/Economics/staff/frainj/home.html
mailto:fra...@tcd.ie
mailto:fra...@gmail.com

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test in R

[fxen3k]

Thanks for your answer, John!

Having read in Wooldridge, Verbeek and Hausman himself, I tried to figure
out how this whole Hausman test works.

I tried to figure out, if endogeneity exists in my particular case. So I did
this

Y ~ X + Z + Rest + error term [# this is the the original regression with Z
= instrumental variable for X, X = potentially endogenous variable and Rest
= more independent variables]
Regression 1:
X ~ Z + Rest + error term
Regression 2:
Y ~ X + Rest + residuals(Reg1) + error [# I took the residuals from
Regression 1 by Reg1_resid - cbind(Red1$resid)

Finally, if the coefficient for the residuals is statistically significant,
there is endogeneity. 

Is this approach correct?

p.s: My p-value is 0.1138...

Thanks for your help





--
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-test-in-R-tp4647716p4647800.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test in R

[Bert Gunter]

1. These are primarily statistics issues, not R issues. You should
post on a statistical help list like stats.stackexchange.com, not
here.

2. However, given your acknowledged statistical ignorance, you may be
asking for trouble. I suggest you seek help from a local statistical
expert to get you started. Then, depending on your statistical
background, you may understand enough to drive safely on your own.

Also try at the R command prompt:

install.packages(fortunes)
library(fortunes)
fortune(brain surgery)

Cheers,
 Bert



On Sun, Oct 28, 2012 at 1:33 PM, fxen3k f.seha...@gmail.com wrote:
 Hi there,

 I am really new to statistics in R and statistics itself as well.
 My situation: I ran a lot of OLS regressions with different independent
 variables. (using the lm() function).
 After having done that, I know there is endogeneity due to omitted
 variables. (or perhaps due to any other reasons).
 And here comes the Hausman test. I know this test is used to identify
 endogeneity.
 But what I am not sure about is: Can I use the Hausman test in a simple OLS
 regression or is it only possible in a 2SLS regression model? And if it is
 possible to use it, how can I do it?

 Info about the data:

 data = lots of data :)

 x1 - data$x1
 x2 - data$x2
 x3 - data$x3
 x4 - data$x4
 y1 - data$y1

 reg1 - summary(lm(y1 ~ x1 + x2 + x3 + x4))

 Thanks in advance for any support!



 --
 View this message in context: 
 http://r.789695.n4.nabble.com/Hausman-test-in-R-tp4647716.html
 Sent from the R help mailing list archive at Nabble.com.

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.



-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test in R

[Joshua Wiley]

Hi,

I can think of no reason a Hausman test could not be used for OLS---it is a
comparison of vectors of coefficients from different models usually assumed
to produce similar estimates under certain conditions.  Dissimilarity is
taken as indicative of a lack of some or all the conditions required for
the two models to yield similar parameters.
I suggest you look at the plm and systemfit packages.  They have many
functions for OLS, 2SLS, tests of endogeneity, etc.  The plm (and maybe
systemfit?) package also has a vignette which is a good thing to read.  It
has a lot of useful information on the code and examples of comparing
different types of models, that you may find instructive.

Hope this helps,

Josh


On Sun, Oct 28, 2012 at 1:33 PM, fxen3k f.seha...@gmail.com wrote:

 Hi there,

 I am really new to statistics in R and statistics itself as well.
 My situation: I ran a lot of OLS regressions with different independent
 variables. (using the lm() function).
 After having done that, I know there is endogeneity due to omitted
 variables. (or perhaps due to any other reasons).
 And here comes the Hausman test. I know this test is used to identify
 endogeneity.
 But what I am not sure about is: Can I use the Hausman test in a simple
 OLS
 regression or is it only possible in a 2SLS regression model? And if it
 is
 possible to use it, how can I do it?

 Info about the data:

 data = lots of data :)

 x1 - data$x1
 x2 - data$x2
 x3 - data$x3
 x4 - data$x4
 y1 - data$y1

 reg1 - summary(lm(y1 ~ x1 + x2 + x3 + x4))

 Thanks in advance for any support!



 --
 View this message in context:
 http://r.789695.n4.nabble.com/Hausman-test-in-R-tp4647716.html
 Sent from the R help mailing list archive at Nabble.com.

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.




-- 
Joshua Wiley
Ph.D. Student, Health Psychology
Programmer Analyst II, Statistical Consulting Group
University of California, Los Angeles
https://joshuawiley.com/

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Achim Zeileis]

On Sun, 16 Jan 2011, Holger Steinmetz wrote:



Hi,

can anybody tell me how the Hausman test for endogenty works?

I have a simulated model with three correlated predictors (X1-X3). I also
have an instrument W for X1

Now I want to test for endogeneity of X1 (i.e., when I omit X2 and X3 from
the equation).

My current approach:

library(systemfit)

fit2sls - systemfit(Y~X1,data=data,method=2SLS,inst=~W)
fitOLS - systemfit(Y~X1,data=data,method=OLS)
print(hausman.systemfit(fitOLS, fit2sls))

This seems to work fine. However, when I include X2 as a furter predictor,
the 2sls-estimation doesn't work.


When you don't need any instruments for X2, then you should employ

  Y ~ X1 + X2, inst = ~ W + X2

Then, regressor X2 is unaltered in the second stage of the regression 
(after projection onto the instruments).


hth,
Z


Thanks in advance
Holger

--
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-Test-tp3220016p3220016.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Holger Steinmetz]

Dear Achim,

thank you very much.

One follow up question. The Hausman-test always gives me a p-value of 1 - no
matter
how small the statistic is.

I now generated orthogonal regressors (X1-X3) and the test gives me


Hausman specification test for consistency of the 3SLS estimation

data:  data 
Hausman = -0.0138, df = 2, p-value = 1

What is confusing to me is the 3SLS. I am just beginning to learn about
instrumental variables (I am a psychologist ;) Perhaps that's a problem?

As a background, here's the complete simulation:

W = rnorm(1000)
X2 = rnorm(1000)
X3 = rnorm(1000)
X1 = .5*W  + rnorm(1000)
Y = .4*X1 + .5*X2 + .6*X3 + rnorm(1000)
data = as.data.frame(cbind(X1,X2,X3,Y,W))

fit2sls - systemfit(Y~X1,data=data,method=2SLS,inst=~W)
fitOLS - systemfit(Y~X1,data=data,method=OLS)

print(hausman.systemfit(fitOLS, fit2sls))

Best,
Holger
-- 
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-Test-tp3220016p3220065.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Arne Henningsen]

Hi Holger!

On 16 January 2011 15:53, Holger Steinmetz holger.steinm...@web.de wrote:
 One follow up question. The Hausman-test always gives me a p-value of 1 - no
 matter how small the statistic is.

 I now generated orthogonal regressors (X1-X3) and the test gives me


        Hausman specification test for consistency of the 3SLS estimation

 data:  data
 Hausman = -0.0138, df = 2, p-value = 1

 What is confusing to me is the 3SLS. I am just beginning to learn about
 instrumental variables (I am a psychologist ;) Perhaps that's a problem?

 As a background, here's the complete simulation:

 W = rnorm(1000)
 X2 = rnorm(1000)
 X3 = rnorm(1000)
 X1 = .5*W  + rnorm(1000)
 Y = .4*X1 + .5*X2 + .6*X3 + rnorm(1000)
 data = as.data.frame(cbind(X1,X2,X3,Y,W))

 fit2sls - systemfit(Y~X1,data=data,method=2SLS,inst=~W)
 fitOLS - systemfit(Y~X1,data=data,method=OLS)

 print(hausman.systemfit(fitOLS, fit2sls))

Please do read the documentation of hausman.systemfit(). I regret that
comparing 2SLS with OLS results has not been implemented yet:

== part of documentation of hausman.systemfit() =
Usage:

hausman.systemfit( results2sls, results3sls )

Arguments:

 results2sls : result of a _2SLS_ (limited information) estimation
  returned by ‘systemfit’.

 results3sls : result of a _3SLS_ (full information) estimation
  returned by ‘systemfit’.

Details:

 The null hypotheses of the test is that all exogenous variables
 are uncorrelated with all disturbance terms.  Under this
 hypothesis both the 2SLS and the 3SLS estimator are consistent but
 only the 3SLS estimator is (asymptotically) efficient.  Under the
 alternative hypothesis the 2SLS estimator is consistent but the
 3SLS estimator is inconsistent.

 The Hausman test statistic is

   m = ( b_2 - b_3 )' ( V_2 - V_3 ) ( b_2 - b_3 )

 where $b_2$ and $V_2$ are the estimated coefficients and their
 variance covariance matrix of a _2SLS_ estimation and $b_3$ and
 $V_3$ are the estimated coefficients and their variance covariance
 matrix of a _3SLS_ estimation.

=

Please don't hesitate to write a new version of hausman.systemfit()
that can also compare 2SLS with OLS results.

Best regards from Copenhagen,
Arne

-- 
Arne Henningsen
http://www.arne-henningsen.name

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Achim Zeileis]

On Sun, 16 Jan 2011, Holger Steinmetz wrote:



Dear Achim,

thank you very much.

One follow up question. The Hausman-test always gives me a p-value of 1 
- no matter how small the statistic is.


I now generated orthogonal regressors (X1-X3) and the test gives me


   Hausman specification test for consistency of the 3SLS estimation

data:  data
Hausman = -0.0138, df = 2, p-value = 1

What is confusing to me is the 3SLS.


Hausman tests can be used for comparisons of various models. The 
implementation in systemfit is intended for comparison of 2SLS and 3SLS 
but can also be (ab)used for comparison of 2SLS and OLS. You just have to 
enter the models in the reverse order, i.e., hausman.systemfit(fit2sls, 
fitOLS).


A worked example that computes the test statistic by hand is also 
included in


  help(Baltagi2002, package = AER)

in the section about the US consumption data, Chapter 11.

An adaptation is also shown below:

  ## data
  library(AER)
  data(USConsump1993, package = AER)
  usc - as.data.frame(USConsump1993)
  usc$investment - usc$income - usc$expenditure

  ## 2SLS via ivreg(), Hausman by hand
  fm_ols - lm(expenditure ~ income, data = usc)
  fm_iv - ivreg(expenditure ~ income | investment, data = usc)
  cf_diff - coef(fm_iv) - coef(fm_ols)
  vc_diff - vcov(fm_iv) - vcov(fm_ols)
  x2_diff - as.vector(t(cf_diff) %*% solve(vc_diff) %*% cf_diff)
  pchisq(x2_diff, df = 2, lower.tail = FALSE)

  ## 2SLS via systemfit(), Hausman via hausman.systemfit()
  library(systemfit)
  sm_ols - systemfit(expenditure ~ income, data = usc, method = OLS)
  sm_iv - systemfit(expenditure ~ income, data = usc, method = 2SLS,
inst = ~ investment)
  hausman.systemfit(sm_iv, sm_ols)

hth,
Z


I am just beginning to learn about
instrumental variables (I am a psychologist ;) Perhaps that's a problem?

As a background, here's the complete simulation:

W = rnorm(1000)
X2 = rnorm(1000)
X3 = rnorm(1000)
X1 = .5*W  + rnorm(1000)
Y = .4*X1 + .5*X2 + .6*X3 + rnorm(1000)
data = as.data.frame(cbind(X1,X2,X3,Y,W))

fit2sls - systemfit(Y~X1,data=data,method=2SLS,inst=~W)
fitOLS - systemfit(Y~X1,data=data,method=OLS)

print(hausman.systemfit(fitOLS, fit2sls))

Best,
Holger
--
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-Test-tp3220016p3220065.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Achim Zeileis]

On Sun, 16 Jan 2011, Arne Henningsen wrote:


Hi Holger!

On 16 January 2011 15:53, Holger Steinmetz holger.steinm...@web.de wrote:

One follow up question. The Hausman-test always gives me a p-value of 1 - no
matter how small the statistic is.

I now generated orthogonal regressors (X1-X3) and the test gives me


       Hausman specification test for consistency of the 3SLS estimation

data:  data
Hausman = -0.0138, df = 2, p-value = 1

What is confusing to me is the 3SLS. I am just beginning to learn about
instrumental variables (I am a psychologist ;) Perhaps that's a problem?

As a background, here's the complete simulation:

W = rnorm(1000)
X2 = rnorm(1000)
X3 = rnorm(1000)
X1 = .5*W  + rnorm(1000)
Y = .4*X1 + .5*X2 + .6*X3 + rnorm(1000)
data = as.data.frame(cbind(X1,X2,X3,Y,W))

fit2sls - systemfit(Y~X1,data=data,method=2SLS,inst=~W)
fitOLS - systemfit(Y~X1,data=data,method=OLS)

print(hausman.systemfit(fitOLS, fit2sls))


Please do read the documentation of hausman.systemfit(). I regret that
comparing 2SLS with OLS results has not been implemented yet:

== part of documentation of hausman.systemfit() =
Usage:

   hausman.systemfit( results2sls, results3sls )

Arguments:

results2sls : result of a _2SLS_ (limited information) estimation
 returned by ?systemfit?.

results3sls : result of a _3SLS_ (full information) estimation
 returned by ?systemfit?.

Details:

The null hypotheses of the test is that all exogenous variables
are uncorrelated with all disturbance terms.  Under this
hypothesis both the 2SLS and the 3SLS estimator are consistent but
only the 3SLS estimator is (asymptotically) efficient.  Under the
alternative hypothesis the 2SLS estimator is consistent but the
3SLS estimator is inconsistent.

The Hausman test statistic is

  m = ( b_2 - b_3 )' ( V_2 - V_3 ) ( b_2 - b_3 )

where $b_2$ and $V_2$ are the estimated coefficients and their
variance covariance matrix of a _2SLS_ estimation and $b_3$ and
$V_3$ are the estimated coefficients and their variance covariance
matrix of a _3SLS_ estimation.

=

Please don't hesitate to write a new version of hausman.systemfit()
that can also compare 2SLS with OLS results.


Arne: Unless I'm missing something, hausman.systemfit() essentially does 
the right thing and computes the right statistic and p-value (see my other 
mail to Holger). Maybe some preliminary check on the input objects could 
be used for determining the right order of models.


Best,
Z


Best regards from Copenhagen,
Arne

--
Arne Henningsen
http://www.arne-henningsen.name

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Arne Henningsen]

Hi Achim!

On 16 January 2011 16:37, Achim Zeileis achim.zeil...@uibk.ac.at wrote:
 Arne: Unless I'm missing something, hausman.systemfit() essentially does the
 right thing and computes the right statistic and p-value (see my other mail
 to Holger). Maybe some preliminary check on the input objects could be used
 for determining the right order of models.

Thanks for the response and the suggestions. Adding a check for the
input objects and extending the documentation is a good idea! I will
change the systemfit package accordingly in the future.

/Arne

-- 
Arne Henningsen
http://www.arne-henningsen.name

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman Test

[Holger Steinmetz]

Thank you both very much !

This helped me a lot.

Best,
Holger
-- 
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-Test-tp3220016p3220123.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test for endogeneity

[Bert Gunter]

... and, in fact, simply googling on R Package Hausmann finds two
Hausmann test functions in 2 different packages within the first half
dozen hits.

-- Bert

On Sat, Oct 9, 2010 at 11:06 AM, Liviu Andronic landronim...@gmail.com wrote:
 Hello

 On Sat, Oct 9, 2010 at 2:37 PM, Holger Steinmetz
 holger.steinm...@web.de wrote:
 can anybody point me in the right direction on how to conduct a hausman test
 for endogeneity in simultanous equation models?

 Try
 install.packages('sos')
 require(sos)
 findFn('hausman')

 Here I get these results:
 findFn('hausman')
 found 22 matches;  retrieving 2 pages
 2

 Liviu

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.




-- 
Bert Gunter
Genentech Nonclinical Biostatistics

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test for endogeneity

[Holger Steinmetz]

Dear Liviu,

thank you very much. After inspecting the options, I *guess* that systemfit
is what I need.
However, I absolutely don't understand how it works. I searched long for a
detailed documentation (beyond the rather cryptic standard documentation)
but found none. 

Has anybody references/advises how to conduct the test?

Best,
Holger
-- 
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-test-for-endogeneity-tp2969522p2970261.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test for endogeneity

[Arne Henningsen]

Hi Holger

On 10 October 2010 15:36, Holger Steinmetz holger.steinm...@web.de wrote:
 After inspecting the options, I *guess* that systemfit
 is what I need.
 However, I absolutely don't understand how it works. I searched long for a
 detailed documentation (beyond the rather cryptic standard documentation)
 but found none.

 Has anybody references/advises how to conduct the test?

A paper describing the systemfit package has been published in the
journal of statistical software:

http://www.jstatsoft.org/v23/i04/paper

It describes the Hausman test for testing the consistency of the 3SLS
estimates against the 2SLS estimates (see sections 2.8 and 4.6).

I guess (but I am not sure -- maybe others can comment on this) that
you test for the endogeneity of regressors, e.g., by

fitSur - systemfit( myFormula, data = myData, method = SUR )

fit3sls - systemfit( myFormula, data = myData, method = 3SLS, inst
= myInst )

hausman.systemfit( fit3sls, fitSur )

If some regressors are endogenous, the SUR estimates are inconsistent
but the 3SLS estimates are consistent given that the instrumental
variables are exogenous. However, if all regressors are exogenous,
both estimates should be consistent but the SUR estimates should be
more efficient.

Best wishes,
Arne

-- 
Arne Henningsen
http://www.arne-henningsen.name

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test for endogeneity

[Holger Steinmetz]

Dear Arne,

this looks promising! Thank you very much.

Best,
Holger
-- 
View this message in context: 
http://r.789695.n4.nabble.com/Hausman-test-for-endogeneity-tp2969522p2970564.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test for endogeneity

[Giuseppe Marinelli]

On Saturday 09 October 2010 14:37:35 Holger Steinmetz wrote:
 Dear folks,

 can anybody point me in the right direction on how to conduct a hausman
 test for endogeneity in simultanous equation models?

 Best,
 Holger

hausman.systemfit [1] should be what you are looking for.
Cheers

Giuseppe

[1] http://cran.r-project.org/web/packages/systemfit/systemfit.pdf

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Hausman test for endogeneity

[Liviu Andronic]

Hello

On Sat, Oct 9, 2010 at 2:37 PM, Holger Steinmetz
holger.steinm...@web.de wrote:
 can anybody point me in the right direction on how to conduct a hausman test
 for endogeneity in simultanous equation models?

Try
install.packages('sos')
require(sos)
findFn('hausman')

Here I get these results:
 findFn('hausman')
found 22 matches;  retrieving 2 pages
2

Liviu

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.