Beruflich Dokumente
Kultur Dokumente
10-1
10-2
10-3
10-4
10-5
16-5
Fixed Effects
Fixed-effects (FE) explore the relationship between the
independent variables and dependent variable within an
entity (country, state, institution etc.).
Each entity (state) has its own individual characteristics that
may or may not influence the dependent variables
Why use FE? Because we believe that something within the
entity (state) will bias the variables; we need to control for
this to get unbiased estimates.
Therefore, FE removes the effect of those time-invariant
characteristics from the independent variables so we can
assess their net effect.
Copyright 2015 Pearson, Inc. All rights reserved.
10-6
Fixed Effects
Fixed effects form
Yit = 1Xit + i + uit
i is called a state fixed effect or state effect it is
the constant (fixed) effect of being in state I
Again, FE removes the effect of those time-invariant
characteristics from the independent variables
thus we CANNOT get a coefficient for a specific time-invariant
variable (race, gender, etc) since they are all lumped in with
the intercept in the term i
10-7
10-8
10-10
.xtsetstateyear;
panelvariable:state(stronglybalanced)
timevariable:year,1982to1988
delta:1unit
10-11
xtregvfrallbeertax,fevce(clusterstate)
Fixed-effects(within)regressionNumberofobs=336
Groupvariable:stateNumberofgroups=48
R-sq:within=0.0407Obspergroup:min=7
between=0.1101avg=7.0
overall=0.0934max=7
F(1,47)=5.05
corr(u_i,Xb)=-0.6885Prob>F=0.0294
(Std.Err.adjustedfor48clustersinstate)
-----------------------------------------------------------------------------|Robust
vfrall|Coef.Std.Err.tP>|t|[95%Conf.Interval]
-------------+---------------------------------------------------------------beertax|-.6558736.2918556-2.250.029-1.243011-.0687358
_cons|2.377075.149796615.870.0002.0757232.678427
------------------------------------------------------------------------------
Thepaneldatacommandxtregwiththeoptionfeperformsfixedeffects
regression.Thereportedinterceptisarbitrary,andtheestimated
individualeffectsarenotreportedinthedefaultoutput.
Thefeoptionmeansusefixedeffectsregression
Thevce(clusterstate)optiontellsSTATAtouseclusteredstandard
errorswhatarethese?Letsfigurethemouttogether
10-12
10-13
10-14
geny83=(year==1983);
geny84=(year==1984);
geny85=(year==1985);
.globalyeardum"y83y84y85y86y87y88";
.xtregvfrallbeertax$yeardum,fevce(clusterstate);
geny86=(year==1986);
geny87=(year==1987);
Fixed-effects(within)regressionNumberofobs=336
geny88=(year==1988);
Groupvariable:stateNumberofgroups=48
R-sq:within=0.0803Obspergroup:min=7
between=0.1101avg=7.0
overall=0.0876max=7
corr(u_i,Xb)=-0.6781Prob>F=0.0009
(Std.Err.adjustedfor48clustersinstate)
-----------------------------------------------------------------------------|Robust
vfrall|Coef.Std.Err.tP>|t|[95%Conf.Interval]
-------------+---------------------------------------------------------------beertax|-.6399799.3570783-1.790.080-1.358329.0783691
y83|-.0799029.0350861-2.280.027-.1504869-.0093188
y84|-.0724206.0438809-1.650.106-.1606975.0158564
y85|-.1239763.0460559-2.690.010-.2166288-.0313238
y86|-.0378645.0570604-0.660.510-.1526552.0769262
y87|-.0509021.0636084-0.800.428-.1788656.0770615
y88|-.0518038.0644023-0.800.425-.1813645.0777568
_cons|2.42847.201688512.040.0002.0227252.834215
-------------+---------------------------------------------------------------Copyright 2015 Pearson, Inc. All rights reserved.
10-15
(1)y83=0
(2)y84=0
(3)y85=0
(4)y86=0
(5)y87=0
(6)y88=0
F(6,47)=4.22
Prob>F=0.0018
Yes
10-16
10-17
10-18
16-18
10-20
16-20
Fixed-effects(within)regressionNumberofobs=336
Groupvariable:stateNumberofgroups=48
R-sq:within=0.0407Obspergroup:min=7
between=0.1101avg=7.0
overall=0.0934max=7
F(1,47)=5.05
corr(u_i,Xb)=-0.6885Prob>F=0.0294
(Std.Err.adjustedfor48clustersinstate)
-----------------------------------------------------------------------------|Robust
vfrall|Coef.Std.Err.tP>|t|[95%Conf.Interval]
-------------+---------------------------------------------------------------beertax|-.6558736.2918556-2.250.029-1.243011-.0687358
_cons|2.377075.149796615.870.0002.0757232.678427
-----------------------------------------------------------------------------Copyright 2015 Pearson, Inc. All rights reserved.
10-21
10-22
Method 2:
hausman alternative command: does the same thing
10-24
10-25
IV Regression: Why?
Three important threats to internal validity are:
Omitted variable bias from a variable that is correlated with
X but is unobserved (so cannot be included in the
regression) and for which there are inadequate control
variables;
Simultaneous causality bias (X causes Y, Y causes X);
Errors-in-variables bias (X is measured with error)
All three problems result in E(u|X) 0.
Instrumental variables regression can eliminate bias when
E(u|X) 0 using an instrumental variable (IV), Z.
10-26
10-27
X i
predicted values
TSLS
. 1
10-28
Pi cigarettes )) 0?
2. Exogenous? corr(SalesTaxi,ui) = 0?
Copyright 2015 Pearson, Inc. All rights reserved.
10-29
Instrumentalvariables(2SLS)regressionNumberofobs=48
Waldchi2(1)=12.05
Prob>chi2=0.0005
R-squared=0.4011
RootMSE=.18635
-----------------------------------------------------------------------------|Robust
lpackpc|Coef.Std.Err.zP>|z|[95%Conf.Interval]
-------------+---------------------------------------------------------------lravgprs|-1.083587.3122035-3.470.001-1.695494-.471679
_cons|9.7198761.4961436.500.0006.7874912.65226
-----------------------------------------------------------------------------Instrumented:lravgprsThis is the endogenous regressor
Instruments:rtaxsoThis is the instrumental varible
-----------------------------------------------------------------------------Estimated
,Pncigarettes
= 48)
ln(
i
10-31
10-32
10-33
10-34
P cigarettes )
Q cigarettes ) ) = + ln(ln(
ln( ln(
) + 2ln(Incomei) + ui
i
0
1
i
We actually have two instruments:
Z1i = general sales taxi
Z2i = cigarette-specific taxi
10-35
Z1
Z2
IV(2SLS)regressionwithrobuststandarderrorsNumberofobs=48
F(2,45)=16.17
Prob>F=0.0000
R-squared=0.4294
RootMSE=.18786
-----------------------------------------------------------------------------|Robust
lpackpc|Coef.Std.Err.tP>|t|[95%Conf.Interval]
-------------+---------------------------------------------------------------lravgprs|-1.277424.2496099-5.120.000-1.780164-.7746837
lperinc|.2804045.25388941.100.275-.230955.7917641
_cons|9.894955.959216910.320.0007.96299311.82692
-----------------------------------------------------------------------------Instrumented: lravgprs
Instruments: lperinc rtaxso rtax STATA lists ALL the exogenous regressors
as instruments slightly different
terminology than we have been using
-----------------------------------------------------------------------------Copyright 2015 Pearson, Inc. All rights reserved.
10-36
(0.37)
(0.31)
10-37
Outline
1. IV Regression: Why and What; Two Stage
Least Squares
2. The General IV Regression Model
3. Checking Instrument Validity
a) Weak and strong instruments
b) Instrument exogeneity
10-38
10-39
10-40
10-41
10-42
X
(Anderson-Rubin test) can only be done if #Zs >
#Xs (overidentified).
Same test works if we have more variables
Yi = 0 + 1X1i + + kXki + k+1W1i + + k+rWri + ui
Copyright 2015 Pearson, Inc. All rights reserved.
10-43
10-44