The Stata Journal

Editor
H. Joseph Newton
Department of Statistics
Texas A & M University
College Station, Texas 77843
979-845-3142; FAX 979-845-3144
[email protected]

Editor
Nicholas J. Cox
Department of Geography
Durham University
South Road
Durham City DH1 3LE UK
[email protected]

Associate Editors
Christopher F. Baum
Boston College
Rino Bellocco
Karolinska Institutet, Sweden and
Univ. degli Studi di Milano-Bicocca, Italy
A. Colin Cameron
University of CaliforniaDavis
David Clayton
Cambridge Inst. for Medical Research
Mario A. Cleves
Univ. of Arkansas for Medical Sciences
William D. Dupont
Vanderbilt University
Charles Franklin
University of WisconsinMadison
Joanne M. Garrett
University of North Carolina
Allan Gregory
Queens University
James Hardin
University of South Carolina
Ben Jann
ETH Z
urich, Switzerland
Stephen Jenkins
University of Essex
Ulrich Kohler
WZB, Berlin

J. Scott Long
Indiana University
Thomas Lumley
University of WashingtonSeattle
Roger Newson
Imperial College, London
Marcello Pagano
Harvard School of Public Health
Sophia Rabe-Hesketh
University of CaliforniaBerkeley
J. Patrick Royston
MRC Clinical Trials Unit, London
Philip Ryan
University of Adelaide
Mark E. Schaffer
Heriot-Watt University, Edinburgh
Jeroen Weesie
Utrecht University
Nicholas J. G. Winter
University of Virginia
Jeffrey Wooldridge
Michigan State University

Stata Press Production Manager

Stata Press Copy Editor

Lisa Gilmore
Gabe Waggoner

Jens Lauritsen
Odense University Hospital
Stanley Lemeshow
Ohio State University

Copyright Statement: The Stata Journal and the contents of the supporting files (programs, datasets, and
c by StataCorp LP. The contents of the supporting files (programs, datasets, and
help files) are copyright
help files) may be copied or reproduced by any means whatsoever, in whole or in part, as long as any copy
or reproduction includes attribution to both (1) the author and (2) the Stata Journal.
The articles appearing in the Stata Journal may be copied or reproduced as printed copies, in whole or in part,
as long as any copy or reproduction includes attribution to both (1) the author and (2) the Stata Journal.
Written permission must be obtained from StataCorp if you wish to make electronic copies of the insertions.
This precludes placing electronic copies of the Stata Journal, in whole or in part, on publicly accessible web
sites, fileservers, or other locations where the copy may be accessed by anyone other than the subscriber.
Users of any of the software, ideas, data, or other materials published in the Stata Journal or the supporting
files understand that such use is made without warranty of any kind, by either the Stata Journal, the author,
or StataCorp. In particular, there is no warranty of fitness of purpose or merchantability, nor for special,
incidental, or consequential damages such as loss of profits. The purpose of the Stata Journal is to promote
free communication among Stata users.
The Stata Journal, electronic version (ISSN 1536-8734) is a publication of Stata Press. Stata and Mata are
registered trademarks of StataCorp LP.

The Stata Journal (2007)

7, Number 2, pp. 197208

Estimation of nonstationary heterogeneous

panels
Edward F. Blackburne III
Sam Houston State University
Huntsville, TX
[email protected]

Mark W. Frank
Sam Houston State University
Huntsville, TX
[email protected]

Abstract. We introduce a new Stata command, xtpmg, for estimating nonstationary heterogeneous panels in which the number of groups and number of time-series
observations are both large. Based on recent advances in the nonstationary panel
literature, xtpmg provides three alternative estimators: a traditional fixed-effects
estimator, the mean-group estimator of Pesaran and Smith (Estimating long-run
relationships from dynamic heterogeneous panels, Journal of Econometrics 68:
79113), and the pooled mean-group estimator of Pesaran, Shin, and Smith (Estimating long-run relationships in dynamic heterogeneous panels, DAE Working
Papers Amalgamated Series 9721; Pooled mean group estimation of dynamic heterogeneous panels, Journal of the American Statistical Association 94: 621634).
Keywords: st0125, xtpmg, nonstationary panels, heterogeneous dynamic panels,
pooled mean-group estimator, mean-group estimator, panel cointegration

Introduction

In recent years, the dynamic panel-data literature has begun to focus on panels in
which the number of cross-sectional observations (N ) and the number of time-series
observations (T ) are both large. The availability of data with greater frequency is
certainly a key contributor to this shift. Some cross-national and cross-state datasets,
for example, are now large enough in T such that each nation (or state) can be estimated
separately.
The asymptotics of large N , large T dynamic panels are different from the asymptotics of traditional large N , small T dynamic panels. Small T panel estimation usually relies on fixed- or random-effects estimators, or a combination of fixed-effects estimators and instrumental-variable estimators, such as the Arellano and Bond (1991)
generalized method-of-moments estimator. These methods require pooling individual
groups and allowing only the intercepts to differ across the groups. One of the central findings from the large N , large T literature, however, is that the assumption of
homogeneity of slope parameters is often inappropriate. This point has been made
by Pesaran and Smith (1995); Im, Pesaran, and Shin (2003); Pesaran, Shin, and Smith
(1997, 1999); and Phillips and Moon (2000).1

1. For more discussion of this literature, see chapter 12 in Baltagi (2001).

c 2007 StataCorp LP

st0125

198

Estimation of nonstationary heterogeneous panels

With the increase in time observations inherent in large N , large T dynamic panels,
nonstationarity is also a concern. Recent papers by Pesaran, Shin, and Smith (1997,
1999) offer two important new techniques to estimate nonstationary dynamic panels in
which the parameters are heterogeneous across groups: the mean-group (MG) and pooled
mean-group (PMG) estimators. The MG estimator (see Pesaran and Smith 1995) relies
on estimating N time-series regressions and averaging the coefficients, whereas the PMG
estimator (see Pesaran, Shin, and Smith 1997, 1999) relies on a combination of pooling
and averaging of coefficients.
In recent empirical research, the MG and PMG estimators have been applied in a
variety of settings. Freeman (2000), for example, uses the estimators to evaluate statelevel alcohol consumption over 19611995. Martinez-Zarzoso and Bengochea-Morancho
(2004) use them in an estimation of an environmental Kuznets curve in a panel of
22 OECD nations over 19751998. Frank (2005) uses the MG and PMG estimators to
evaluate the long-term effect of income inequality on economic growth in a panel of U.S.
states over 19452001.

The MG and PMG estimators

Assume an autoregressive distributive lag (ARDL) (p, q1 , . . . , qk ) dynamic panel specification of the form
yit =

p
X

ij yi,tj +

q
X

ij Xi,tj + i + it

(1)

j=0

j=1

where the number of groups i = 1, 2, . . . , N ; the number of periods t = 1, 2, . . . , T ; X it

is a k 1 vector of explanatory variables; it are the k 1 coefficient vectors; ij are
scalars; and i is the group-specific effect. T must be large enough such that the model
can be fitted for each group separately. Time trends and other fixed regressors may be
included.
If the variables in (1) are, for example, I(1) and cointegrated, then the error term
is an I(0) process for all i. A principal feature of cointegrated variables is their responsiveness to any deviation from long-run equilibrium. This feature implies an error
correction model in which the short-run dynamics of the variables in the system are
influenced by the deviation from equilibrium. Thus it is common to reparameterize (1)
into the error correction equation
0

yit = i (yi,t1 i Xit ) +

p1
X
j=1

ij yi,t1 +

q1
X

ij Xi,tj + i + it

(2)

j=0

Pp
Pq
P
Pp
where i = (1 j=1 ij ), i = j=0 ij /(1 k ik ), ij = m=j+1 im
Pq

1, 2, . . . , p 1, and ij
= m=j+1 im j = 1, 2, . . . , q 1.

j=

The parameter i is the error-correcting speed of adjustment term. If i = 0, then

there would be no evidence for a long-run relationship. This parameter is expected to

E. F. Blackburne and M. W. Frank

199

be significantly negative under the prior assumption that the variables show a return
0
to a long-run equilibrium. Of particular importance is the vector i , which contains the
long-run relationships between the variables.
The recent literature on dynamic heterogeneous panel estimation in which both N
and T are large suggests several approaches to the estimation of (2). On one extreme,
a fixed-effects (FE) estimation approach could be used in which the time-series data for
each group are pooled and only the intercepts are allowed to differ across groups. If
the slope coefficients are in fact not identical, however, then the FE approach produces
inconsistent and potentially misleading results. On the other extreme, the model could
be fitted separately for each group, and a simple arithmetic average of the coefficients
could be calculated. This is the MG estimator proposed by Pesaran and Smith (1995).
With this estimator, the intercepts, slope coefficients, and error variances are all allowed
to differ across groups.
More recently, Pesaran, Shin, and Smith (1997, 1999) have proposed a PMG estimator that combines both pooling and averaging. This intermediate estimator allows the
intercept, short-run coefficients, and error variances to differ across the groups (as would
the MG estimator) but constrains the long-run coefficients to be equal across groups (as
would the FE estimator). Since (2) is nonlinear in the parameters, Pesaran, Shin, and
Smith (1999) develop a maximum likelihood method to estimate the parameters.
Expressing the likelihood as the product of each cross-sections likelihood and taking
the log yields
0

lT ( , , ) =

N
N
1X 1
T X
ln(2i2 )
{yi i i ()}0 Hi {yi i i ()}
2 i=1
2 i=1 i2

(3)

for i = 1, . . . , N , where i () = yi,t1 Xi i , Hi = IT Wi (Wi0 Wi )Wi , IT is an identity

matrix of order T , and Wi = (yi,t1 , . . . , yi,tp+1 , Xi , Xi,t1 , . . . , Xi,tq+1 ).
xtpmg uses Statas powerful ml framework to implement the PMG estimator. Specifically, we take advantage of the undocumented hold option of ml to maximize the
likelihood via back-substitution.2 Beginning with an initial estimate of the long-run
b the short-run coefficients and the group-specific speed of adjustment
coefficient vector, ,
terms can be estimated by regressions of yi on (bi , Wi ). These conditional estimates
are in turn used to update the estimate of . The process is iterated until convergence
is achieved.
The parameter estimates from iterated conditional likelihood maximization are asymptotically identical to those from full-information maximum likelihood. But the
estimated covariance matrix is not. However, since the distribution of the PMG parameters is known, we can recover the full covariance matrix for all estimated parameters.
As shown in Pesaran, Shin, and Smith (1999), the covariance matrix can be estimated
by the inverse of
2. Although (3) looks benign, the model is difficult to program directly as a Stata ml program.
Looking forward to (8), readers will note that the model cannot be readily adapted to mls familiar
theta notation.

200

Estimation of nonstationary heterogeneous panels

P
N
i=1

b2 X 0 Xi

i
i

bi2

b1 X 0 b1

b12
0 b
b
1 1

b12

bN X 0 bN

N
2

bN

..
.

0
..
.

b1 X 0 W1

b12
0
b
1 W1

b12

0 b
bN
N
2

bN

bN X 0 WN

N
2

bN

..
.

..
.

0
..
.

0
bN
WN
2

bN

..
.

0
..
.

W10 W1

b12

0
WN
WN
2

bN

The MG parameters are simply the unweighted means of the individual coefficients.
For example, the MG estimate of the error correction coefficient, , is
b = N 1

with the variance

N
X
i=1

bi

(4)

bb=

X
1
b2
(bi )
N (N 1) i=1

(5)

The mean and variance of other short-run coefficients are similarly estimated.

The xtpmg command

3.1

Syntax

xtpmg varlist

if

 

in

 

, lr(varlist) ec(string) replace constraints(string)

noconstant cluster(varname) level(#) technique(algorithm spec)


difficult full model

3.2

Options

lr(varlist) specifies the variables to be included when calculating the long-run cointegrating vector.
ec(string) is used to specify the name of the newly created error-correction term; default
is ec.
replace overwrites the error-correction variable, if it exists.
constraints(string) specifies the constraints to be applied to the model. This option
is currently used only with option pmg.

E. F. Blackburne and M. W. Frank

201

noconstant suppresses the constant term. This option cannot be used with option dfe.
cluster(varname) specifies that the observations are independent across groups (clusters), but not necessarily within groups. varname specifies to which group each
observation belongs, e.g., cluster(personid) in data with repeated observations
on individuals. cluster() affects the estimated standard errors and variance
covariance matrix of the estimators (VCE), but not the estimated coefficients; see
[U] 20.14 Obtaining robust variance estimates.
level(#) sets the confidence level; default is level(95).
algorithm spec is
technique(algorithm

 spec) specifies
   the ml
 maximization technique.

algorithm # algorithm #
. . . , where algorithm is nr | bfgs | dfp . The
bhhh algorithm is not compatible with xtpmg. technique() can be used only with
option pmg.
difficult will use a different stepping algorithm in nonconcave regions of the likelihood.
full specifies that all N cross-section regression results be listed. Only the averaged
coefficients are listed by default.
model is the type of estimator to be fitted and is one of the following:
pmg is the default and specifies the PMG estimator. This model constrains the longrun coefficient vector to be equal across panels while allowing for group-specific
short-run and adjustment coefficients.
mg specifies the MG estimator. This model fits parameters as averages of the N
individual group regressions.
dfe specifies the dynamic fixed-effects estimator.

4
4.1

Empirical example: OECD consumption

Data

We illustrate the use of xtpmg with annual aggregate consumption data for 24 Organisation for Economic Co-operation and Development (OECD) nations. These data are taken
from Pesaran, Shin, and Smith (1997, 1999) and encompass the years 19601993.3 The
1993 annual observation for Belgium is not included in the estimation sample, leaving
an estimation period of 19621992 for Belgium and 19621993 for the other 23 OECD
countries. xtpmg requires that the data be tsset before estimation.
. use jasa2
. tsset id year
panel variable:
time variable:

id (unbalanced)
year, 1960 to 1993

3. The original data and GAUSS code are available on Pesarans web site:
http://www.econ.cam.ac.uk/faculty/pesaran.

202

Estimation of nonstationary heterogeneous panels

Assume the long-run consumption function
cit = 0t + 1t yit + 2t it + i + it

(6)

where the number of nations i = 1, 2, . . . , N ; the number of periods t = 1, 2, . . . , T ; c it

is the log of real per capita consumption; yit is the log of real per capita income; and
it is the inflation rate. If the variables are I(1) and cointegrated, then the error term
is I(0) for all i. The ARDL(1,1,1) dynamic panel specification of (6) is
cit = 10i yit + 11i yi,t1 + 20i it + 21i i,t1 + i ci,t1 i + it

(7)

The error correction reparameterization of (7) is

cit = i (ci,t1 0i 1i yit 2i it ) + 11i yit + 21i it + it
where i = (1 i ), 0i =

i
1i ,

it =

10i +11i
1i ,

and 2i =

(8)

20i +21i
1i .

The error-correction speed of adjustment parameter, i , and the long-run coefficients, 1i and 2i , are of primary interest. With the inclusion of 0i , a nonzero mean
of the cointegrating relationship is allowed. One would expect i to be negative if the
variables exhibit a return to long-run equilibrium. Most aggregate consumption theories
indicate that the long-run income elasticity, 1i , should be equal to one. The inflation
effect, 2i , is generally thought to be negative.

4.2

PMG estimation

The first example estimates the PMG estimator for model (8). In this context, the PMG
model allows for heterogeneous short-run dynamics and common long-run income and
inflation elasticities. Often only the long-run parameters are of interest. The default
results of the pmg option include the long-run parameter estimates and the averaged
short-run parameter estimates.4

4. The PMG standard errors match the GAUSS output provided by Pesaran, Shin, and Smith (1999).
The standard errors in table 1 of Pesaran, Shin, and Smith (1999) are, however, different from those
reported here and from the original GAUSS program.

E. F. Blackburne and M. W. Frank

203

. xtpmg d.c d.pi d.y if year>=1962, lr(l.c pi y) ec(ec) replace pmg

Iteration 0:
log likelihood = 2270.3017 (not concave)
Iteration 1:
log likelihood = 2319.1636
Iteration 2:
log likelihood = 2322.9301
Iteration 3:
log likelihood = 2326.7546
Iteration 4:
log likelihood = 2327.0742
Iteration 5:
log likelihood = 2327.0749
Iteration 6:
log likelihood = 2327.0749
Pooled Mean Group Regression
(Estimate results saved as pmg)
Panel Variable (i): id
Time Variable (t): year

Number of obs
Number of groups
Obs per group: min
avg
max
Log Likelihood

Coef.

Std. Err.

P>|z|

=
=
=
=
=
=

767
24
31
32.0
32
2327.075

[95% Conf. Interval]

ec
pi
y

-.4658438
.9044336

.0567332
.0086815

-8.21
104.18

0.000
0.000

-.5770388
.8874181

-.3546487
.9214491

ec
pi
D1.
y
D1.
_cons

-.1998761

.0321683

-6.21

0.000

-.2629247

-.1368275

-.0182588

.0277523

-0.66

0.511

-.0726522

.0361347

.3269355
.1544507

.0574236
.0216943

5.69
7.12

0.000
0.000

.2143873
.1119308

.4394838
.1969707

SR

In the output, the estimated long-run inflation elasticity is significantly negative, as

expected. Also, the estimated income elasticity is significantly positive. Theoretically,
the income elasticity is equal to one. This hypothesis is easily tested:
. test [ec]y=1
( 1)

[ec]y = 1
chi2( 1) =
Prob > chi2 =

121.18
0.0000

The corresponding 2 value of 121.2 leads to rejection of the null hypothesis of unit
income elasticity.
The full option estimates and saves an N + 1 multiple-equation model. The first
equation (labeled per option ec) presents the normalized cointegrating vector. 5 The
remaining N equations list the group-specific short-run coefficients.

5. The vector has been normalized such that the coefficient on the first term in the cointegrating
vector is 1. Accordingly, the normalized term is omitted from the estimation output.

204

Estimation of nonstationary heterogeneous panels

. xtpmg d.c d.pi d.y if year>=1962, lr(l.c pi y) ec(ec) full pmg
Iteration 0:
log likelihood = 2270.3017 (not concave)
Iteration 1:
log likelihood = 2319.1636
Iteration 2:
log likelihood = 2322.9301
Iteration 3:
log likelihood = 2326.7546
Iteration 4:
log likelihood = 2327.0742
Iteration 5:
log likelihood = 2327.0749
Iteration 6:
log likelihood = 2327.0749
Pooled Mean Group Regression
(Estimate results saved as PMG)
Panel Variable (i): id
Time Variable (t): year

Coef.

Number of obs
Number of groups
Obs per group: min
avg
max
Log Likelihood
Std. Err.

P>|z|

=
=
=
=
=
=

767
24
31
32.0
32
2327.075

[95% Conf. Interval]

ec
pi
y

-.4658438
.9044336

.0567332
.0086815

-8.21
104.18

0.000
0.000

-.5770388
.8874181

-.3546487
.9214491

ec
-.0378815
pi
D1.
-.2114431
y
D1.
.5195067
_cons
.0336383
(output omitted )
id_196
ec
-.4978606
pi
D1.
.0721044
y
D1.
.0390557
.2743539
_cons

.0240594

-1.57

0.115

-.0850371

.0092742

.0866913

-2.44

0.015

-.3813548

-.0415314

.055876
.0147912

9.30
2.27

0.000
0.023

.4099918
.0046481

.6290217
.0626285

.0887771

-5.61

0.000

-.6718605

-.3238608

.0721146

1.00

0.317

-.0692375

.2134464

.103316
.0630399

0.38
4.35

0.705
0.000

-.16344
.1507979

.2415515
.3979099

id_111

Since each group has its own estimated equation, we can, for example, predict variables intuitively.
. predict dc111 if id==111, eq(id_111)
(783 missing values generated)

Similarly, cross-equation restrictions are easily applied.

. test [id_111]ec=[id_112]ec=0
( 1) [id_111]ec - [id_112]ec = 0
( 2) [id_111]ec = 0
chi2( 2) =
2.54
Prob > chi2 =
0.2814

E. F. Blackburne and M. W. Frank

4.3

205

MG estimation

The MG estimates are the unweighted mean of the N individual regression coefficients.
xtpmg with the mg option loops through all panels in the sample to estimate the parameters of (8).6
. xtpmg d.c d.pi d.y if year>=1962, lr(l.c pi y) ec(ec) replace mg
Mean Group Estimation: Error Correction Form
(Estimate results saved as mg)
Coef.

Std. Err.

P>|z|

[95% Conf. Interval]

ec
pi
y

-.3529095
.9181344

.1168025
.0272673

-3.02
33.67

0.003
0.000

-.5818381
.8646914

-.1239809
.9715774

ec
pi
D1.
y
D1.
_cons

-.3063473

.0301599

-10.16

0.000

-.3654597

-.2472349

-.0253642

.0294774

-0.86

0.390

-.0831389

.0324104

.2337588
.2082185

.0489502
.1089385

4.78
1.91

0.000
0.056

.1378182
-.005297

.3296994
.4217339

SR

The MG estimates are presented as a two-equation model: the normalized cointegrating vector and the short-run dynamic coefficients. In comparing the PMG and MG
estimators, we note that the estimated long-run income and inflation elasticities are
statistically significant and properly signed in both models. However, the PMG estimate of the inflation elasticity is larger in magnitude than the estimate from the MG
model (.47 and .35, respectively). The opposite is true for the estimated long-run
income elasticity (.90 and .92, respectively). The speed of adjustment estimates from
each model imply significantly different short-run dynamics (compare b = .20 from
b = .31 from MG).
PMG and
Recall that the PMG estimator constrains the long-run elasticities to be equal across
all panels. This pooling across countries yields efficient and consistent estimates when
the restrictions are true. Often, however, the hypothesis of slope homogeneity is rejected
empirically. If the true model is heterogeneous, the PMG estimates are inconsistent; the
MG estimates are consistent in either case. The test of difference in these models is
performed with the familiar Hausman test.7

6. Actually, since (8) is nonlinear in the parameters, xtpmg estimates the reduced-form regressions for
each group, ct = ct1 + 1 yt + 2 t + 1 yt + 2 t , and then applies Statas nlcom command to
recover the underlying parameter estimates.
7. Statas hausman test offers a sigmamore option. This option forces the variancecovariance matrix
from the efficient model (PMG here) to be used in calculating the test statistic. This is what is presented
here. See Baum, Schaffer, and Stillman (2003) for more details.

206

Estimation of nonstationary heterogeneous panels

. hausman mg pmg, sigmamore
Coefficients
(b)
(B)
mg
pmg
pi
y

Test:

-.3529095
.9181344

(b-B)
Difference

-.4658438
.9044336

.1129342
.0137008

sqrt(diag(V_b-V_B))
S.E.
.126218
.0311167

b = consistent under Ho and Ha; obtained from xtpmg

B = inconsistent under Ha, efficient under Ho; obtained from xtpmg
Ho: difference in coefficients not systematic
chi2(2) = (b-B)[(V_b-V_B)^(-1)](b-B)
=
1.06
Prob>chi2 =
0.5887

The calculated Hausman statistic is 1.06 and is distributed 2 (2). Here we conclude
that the PMG estimator, the efficient estimator under the null hypothesis, is preferred.

4.4

Dynamic FE

The dynamic FE estimator, like the PMG estimator, restricts the coefficients of the cointegrating vector to be equal across all panels. The FE model further restricts the speed
of adjustment coefficient and the short-run coefficients to be equal. xtpmg with the dfe
option fits the model in (8) while allowing panel-specific intercepts.8 An allowance for
intragroup correlation in the calculation of standard errors is made with the cluster()
option.
. xtpmg d.c d.pi d.y if year>=1962, lr(l.c pi y) ec(ec) replace dfe cluster(id)
Standard errors adjusted with cluster(id) option.
Dynamic Fixed Effects Regression: Estimated Error Correction Form
(Estimate results saved as DFE)
Coef.

Std. Err.

P>|z|

[95% Conf. Interval]

ec
pi
y

-.266343
.9120574

.102506
.0468008

-2.60
19.49

0.009
0.000

-.4672509
.8203295

-.065435
1.003785

ec
pi
D1.
y
D1.
_cons

-.1794146

.0434584

-4.13

0.000

-.2645915

-.0942378

-.0280826

.0325622

-0.86

0.388

-.0919034

.0357382

.3811944
.1257634

.070876
.0805454

5.38
1.56

0.000
0.118

.24228
-.0321025

.5201089
.2836294

SR

8. For the FE model, xtpmg is simply a wrapper for Statas xtreg, fe command (see [XT] xtreg).
The underlying parameters of (8) are calculated with nlcom and stored as EC. The reduced-form model,
as estimated by xtreg, fe, is stored as DFE.

E. F. Blackburne and M. W. Frank

207

All coefficients from the dynamic FE model are properly signed and, in fact, similar
to the PMG and MG estimates. As discussed in Baltagi, Griffin, and Xiong (2000),
FE models are subject to a simultaneous equation bias from the endogeneity between
the error term and the lagged dependent variable. The Hausman test can be easily
performed to measure the extent of this endogeneity.
. hausman mg DFE, sigmamore
Coefficients
(b)
(B)
mg
DFE
pi
y

Test:

-.3529095
.9181344

-.266343
.9120574

(b-B)
Difference

sqrt(diag(V_b-V_B))
S.E.

-.0865666
.0060771

25.80672
6.024402

b = consistent under Ho and Ha; obtained from xtpmg

B = inconsistent under Ha, efficient under Ho; obtained from xtpmg
Ho: difference in coefficients not systematic
chi2(2) = (b-B)[(V_b-V_B)^(-1)](b-B)
=
0.00
Prob>chi2 =
1.0000

Results indicate that the simultaneous equation bias is minimal for these data and, for
this example, we conclude that the FE model is preferred over the MG model.

Conclusion

This paper follows recent advances offered by Pesaran and Smith (1995) and Pesaran,
Shin, and Smith (1997; 1999) in the estimation of nonstationary heterogeneous panels
with large N and large T . We offer a new Stata command, xtpmg, that estimates
three alternative models: a traditional dynamic FE estimator that relies on pooling of
cross-sections, an MG estimator that relies on averaging of cross-sections, and a PMG
estimator that relies on a combination of pooling and averaging of coefficients.

References

Arellano, M., and S. Bond. 1991. Some tests of specification for panel data: Monte
Carlo evidence and an application to employment equations. Review of Economic
Studies 58: 277297.
Baltagi, B. H. 2001. Econometric Analysis of Panel Data. 2nd ed. New York: Wiley.
Baltagi, B. H., J. M. Griffin, and W. Xiong. 2000. To pool or not to pool: Homogeneous
versus heterogeneous estimators applied to cigarette demand. Review of Economics
and Statistics 82: 117126.
Baum, C. F., M. E. Schaffer, and S. Stillman. 2003. Instrumental variables and GMM:
Estimation and testing. Stata Journal 3: 131.

208

Estimation of nonstationary heterogeneous panels

Frank, M. W. 2005. Income inequality and economic growth in the U.S.: A panel
cointegration approach. Sam Houston State University Working Paper 05-03.
Freeman, D. G. 2000. Alternative panel estimates of alcohol demand, taxation, and the
business cycle. Southern Economic Journal 67: 325344.
Im, K. S., M. H. Pesaran, and Y. Shin. 2003. Testing for unit roots in heterogeneous
panels. Journal of Econometrics 115: 5374.
Martinez-Zarzoso, I., and A. Bengochea-Morancho. 2004. Pooled mean group estimation
of an environmental kuznets curve for CO2 . Economics Letters 82: 121126.
Pesaran, M. H., Y. Shin, and R. P. Smith. 1997. Estimating long-run relationships in
dynamic heterogeneous panels. DAE Working Papers Amalgamated Series 9721.
. 1999. Pooled mean group estimation of dynamic heterogeneous panels. Journal
of the American Statistical Association 94: 621634.
Pesaran, M. H., and R. P. Smith. 1995. Estimating long-run relationships from dynamic
heterogeneous panels. Journal of Econometrics 68: 79113.
Phillips, P. C. B., and H. R. Moon. 2000. Nonstationary panel data analysis: An
overview of some recent developments. Econometric Reviews 19: 263286.
About the authors
Edward F. Blackburne III is an associate professor in the Department of Economics and International Business, Sam Houston State University. His current research interests are benchmarking
racial profiling and modeling asymmetric time series.
Mark W. Frank is an associate professor in the Department of Economics and International
Business, Sam Houston State University. His current research interests are measuring effects
of income inequality on economic growth and the economics of alcohol advertising.