Panel Data Analysis

Introduction

 Model Representation
 N-first or T-first representation
 Pooled Model
 Fixed Effects Model
 Random Effects Model
 Asymptotic Theory
 N→∞, or T→∞
 N→∞, T→∞
 Panel-Robust Inference
Panel Data Analysis
Introduction

 The Model
yit  xit'    it 
  it   ui  vt  eit
'
 y x
 it  ui  vt  eit 
it

 One-Way (Individual) Effects: yit  xit   ui  eit

'

 Unobserved Heterogeneity
 Cross Section and Time Series Correlation

Cov(ui , u j )  0, Cov(eit , e jt )  0, i  j
Cov(eit , ei )  0, t  
Panel Data Analysis
Introduction

 N-first Representation  T-first Representation

yit  xit' β  ui  eit yti  xti' β  ui  eti
i  1, 2,..., N ; t  1, 2,..., T t  1, 2,..., T ; i  1, 2,..., N
 
y i  Xi β  ui iT  ei y t  Xt β  u  et
 
y  Xβ  (I N  iT )u  e y  Xβ  (iT  I N )u  e

 Dummy Variables y  Xβ  Du  e
Representation D  I N  iT or D  iT  I N
Panel Data Analysis
Introduction

 Notations
 yi1   xi' 1   x1,i1 x2,i1 xK ,i1   ei1   1 
y   '  x x2,i 2 xK ,i 2  e   
x
y i   i 2  , Xi   i 2    , ei   i 2  , β   2 
1,i 2

         
   '       
 iT 
y  xiT   x1,iT x2,iT xK ,iT   iT 
e K 

 yt1   xt' 1   x1,t1 x2,t1 xK , t 1   et1   u1 

y   '  x x2,t 2 xK ,t 2  e  u 
x
y t   t 2  , Xt   t 2    , et   t 2  , u   2 
1,t 2

         
   '       
 ytN   xtN   x1,tN x2,tN xK ,tN   tN 
e u N 
Example: Investment Demand
 Grunfeld and Griliches [1960]
Iit  i   Fit   Cit   it

 i = 10 firms: GM, CH, GE, WE, US, AF, DM, GY,

UN, IBM; t = 20 years: 1935-1954
 Iit = Gross investment
 Fit = Market value
 Cit = Value of the stock of plant and equipment
Pooled (Constant Effects) Model

yit  xit' β  ui  eit (i  1, 2,..., N ; t  1, 2,..., T )

 assuming u  ui i
yit  xit' β  u  eit , or
β 
yit   x 1    eit  y  Xβ  e
'
it
u 
E (e | X)  0, Var (e | X)   e2I
Fixed Effects Model
yit  xit' β  ui  eit (i  1, 2,..., N ; t  1, 2,..., T )
 ui is fixed, independent of eit, and may be
correlated with xit.
Cov(ui , eit )  0, Cov(ui , xit )  0
y i  Xi   ui iT  ei , i  1, 2,..., N

y t  Xt   u  et , t  1, 2,..., T
Fixed Effects Model
 Fixed Effects Model
 Classical Assumptions
 Strict Exogeneity: E (eit | u, X)  0
 Homoschedasticity: Var (eit | u, X)   e2
 No cross section and time series correlation:
Var (e | u, X)   e2I NT

 Extensions: Var (e | u, X)  
 Panel Robust Variance-Covariance Matrix
Random Effects Model
 Error Components
yit  xit' β   it
 it  ui  eit (i  1, 2,..., N ; t  1, 2,..., T )
 ui is random, independent of eit and xit.
Cov(ui , eit )  0, Cov(ui , xit )  0, Cov(eit , xit )  0
 Define the error components as it = ui + eit
y i  Xi   (uiiT  ei ), i  1, 2,..., N

y t  Xt   (u  et ), t  1, 2,..., T
Random Effects Model

 Random Effects Model

 Classical Assumptions
 Strict Exogeneity
E (eit | X)  0, E (ui | X)  0  E ( it | X)  0
 X includes a constant term, otherwise E(ui|X)=u.
 Homoschedasticity
Var (eit | X)   e2 , Var (ui | X)   u2 , Cov(ui , eit )  0
 Var ( it | X)   2   e2   u2
 Constant Auto-covariance (within panels)
Var (εi | X)   e2IT   u2iT iT'
Random Effects Model

 Random Effects Model

 Classical Assumptions (Continued)
 Cross Section Independence
Var (εi | X)     e2IT   u2iT iT'
Var (ε | X)  Ω  I N  

 Extensions:
 Panel Robust Variance-Covariance Matrix
Fixed Effects Model
Estimation

 Within Model Representation

yit  xit' β  ui  eit
yi  xi' β  ui  ei
yit  yi  (xit'  xi' )β  (eit  ei )
 yit  xit' β  eit
y i  X i β  ei or
Qy i  QXi β  Qei
1 '
where Q  IT  iT iT , (QiT  0, Q ' Q  Q)
T
Fixed Effects Model
Estimation

 Model Assumptions
E (eit | xit )  0
Var (eit | xit )  (1  1/ T ) e2
Cov(eit , eis | xit , xis )  (1/ T ) e2  0, t  s

1 '
Var (ei | Xi )     Q   (IT  iT iT )
2
e
2
e
T
Var (e | X)  Ω  I N  
Fixed Effects Model
Estimation: OLS

 Within Estimator: OLS

y i  X i β  ei  y  Xβ  e
1

  i 1 i i   i 1 i y i
ˆβ  ( X' X) 1 X' y   N X' X  X
N'
OLS

ˆ (βˆ )  ( X' X) 1 X'ΩX( X' X) 1

Var OLS
1 1
 ˆ   i 1 X Xi    i 1 X QXi    i 1 X Xi 
2 N ' N ' N '
e i
  
i

i

1
 ˆ   i 1 X Xi 
2 N '
e
 i

ˆ e2  eˆ ' eˆ / ( NT  N  K ), eˆ  y  Xβˆ
Fixed Effects Model
Estimation: ML

 Normality Assumption
yit  xit' β  ui  eit (t  1, 2,..., T )
y i  Xi β  ui iTi  ei (i  1, 2,..., N )
ei ~ normal iid (0,  e2 IT )

y i  Xi β  ei with y i  Qy i , Xi  QXi , ei  Qei ,
1 '
Q  IT  iT iT
T
ei ~ normal (0, ), where    e2QQ '   e2Q
Fixed Effects Model
Estimation: ML

 Log-Likelihood Function
T 1 1 ' 1
lli (β,  | y i , Xi )   ln  2   ln   ei  ei
2
e
2 2 2
T T 1 1 ' 1
  ln  2   ln( e )  ln Q  2 ei Q ei
2

2 2 2 2 e
 Since Q is singular and |Q|=0, we maximize
T T 1 '
lli (β,  | y i , Xi )   ln  2   ln( e )  2 ei ei
2 2

2 e
e
2 2
Fixed Effects Model
Estimation: ML

 ML Estimator
(β,  e ) ML  arg max  i 1 lli (β,  e2 | y i , Xi )
ˆ 2 N

 i1 i i
ˆ ' eˆ
N
e  1 2 ˆ
ˆ 
2
e  1   ˆ e , ei  y i  Xi βˆ
NT  T
 T  ˆ
e ' ˆ
e
 ˆ e2   
 e
ˆ 2

 T 1  N (T  1)
Fixed Effects Model
Hypothesis Testing

 Pool or Not Pool

yit  xit' β  ui  eit
 F-Test based on dummy
variable model: constant or vs. (ui  u, i )
zero coefficients for D w.r.t
F(N-1,NT-N-K)
yit  xit' β  u  eit
 F-test based on fixed
effects (unrestricted) model
vs. pooled (restricted)
model
( RSS R  RSSUR ) / N  1
F ~ F ( N  1, NT  N  K )
RSSUR / ( NT  N  K )
RSSUR  eˆ 'FE eˆ FE , RSS R  eˆ 'PO eˆ PO
Random Effects Model
Estimation: GLS

 The Model
y i  Xi β  ε i , ε i  ui iT  ei
E (ε i | Xi )  0
Var (ε i | Xi )     e2 IT   u2 iT iT'
  2
 T  2

  e Q 
2 e u
 IT  Q  
 e2

1 ' 1 '
where Q  IT  iT iT , IT  Q  iT iT
T T
Random Effects Model
Estimation: GLS

 GLS
1
βˆ GLS  ( XΩ X) XΩ y    i 1 Xi  Xi 
1 1 1 1
 i 1 i y i
1
N N
X 
 
1
Var (βˆ GLS )  ( XΩ X)    i 1 Xi  Xi 
1 1 1
N

 
1   2
'  1   e2 
2  T
1
where   2 IT  2 u
i i  2
2 T T Q  2 I  Q 
e   e  T u  e   e  T u 

1
1   e2 
2  T
and  2
 Q  I  Q 
 e   e  T u
2

Random Effects Model
Estimation: GLS

 Feasible GLS
 Based on estimated residuals of fixed effects
model ˆ 2  eˆ ' eˆ / N (T  1)
e

1 T ˆ
ˆ  T ˆ  ˆ  T e ' e / N , where ei   t 1 eit
2
1
2
u
2
e
ˆ ˆ ˆ
T
βˆ GLS  ( XΩ
ˆ 1X)1 XΩˆ 1y

Var (βˆ )  ( XΩ
GLS
ˆ 1X) 1
 1 1  2
where    2 Q  2  IT  Q   , ˆ1  ˆ e2  Tˆ u2
ˆ 1

 ˆ e ˆ1 
Random Effects Model
Estimation: GLS

 Feasible GLS
 Within Model Representation

 e2  e2
  1  1
1 2
T  u2   e2
yit   yi  (xit'   xi' )   ( it   i )
yit  xit'    it
 it   it   i  (1   )ui  (eit   ei )
E ( it )  0, Var ( it )   e2
Cov( it ,  i )  Cov( it ,  jt )  0
Random Effects Model
Estimation: ML

 Log-Likelihood Function

yit  xit' β  (ui  eit )  xit' β   it (t  1, 2,..., T )

y i  Xi β  ε i (i  1, 2,..., N )
εi ~ normal iid (0, )

T 1 1
lli (β,  e2 ,  u2 | y i , Xi )   ln  2   ln   εi  1εi
2 2 2
Random Effects Model
Estimation: ML

 where
  e2  T  u2 
   I   i i  Q 
2 2 '
(I T  Q ) 
e T u T T
 e2

1 1   2
 1   2

  2  IT  u
i i  2 Q 
'
2 T T 
e
(I T  Q ) 
e  T u   e
2
 e  T u   e
2 2

 2
 T  2

|  | ( e ) IT  2 iT iT  ( e ) 1  2 
2 T u ' 2 T u
e  e 
Random Effects Model
Estimation: ML

 ML Estimator
(βˆ , ˆ e2 , ˆ u2 ) ML  arg max  i 1 lli (β,  e2 ,  u2 | y i , Xi )
N

where
T 1 1
lli (β,  ,  | y i , Xi )   ln  2   ln   εi  1εi
2
e
2
u
2 2 2
  2
  2

T

  ln 2 e  ln 
2
1
2   e2
e T u

2 
1  T  2 2
 2   t 1 ( yit  xit β)  2
2   ( yit  xit β)  
' u T '

2 e     e  T  u  t 1
2  
Random Effects Model
Hypothesis Testing

 Pool or Not Pool

 Test for Var(ui) = 0, that is
Cov( it , is )  Cov(ui  eit ,ui  eis )  Cov(eit ,eis )
 For balanced panel data, the Lagrange-multiplier
test statistic (Breusch-Pagan, 1980) is:
Random Effects Model
Hypothesis Testing

 Pool or Not Pool (Cont.)

NT  eˆ '(J T  I N )eˆ 
LM    1 ~  2
(1)
2 T  1  eˆ ' eˆ 

 
2
 2

NT   i 1  t 1 it
N T
ˆ
e 
   1
2 T  1   eˆit N T 2
 i 1 t 1 
 
βˆ 
where eˆit  yit   xit 1  
'

uˆ  Pooled
Random Effects Model
Hypothesis Testing

 Fixed Effects vs. Random Effects

H 0 : Cov(ui , xit' )  0 (random effects)
H1 : Cov(ui , xit' )  0 ( fixed effects)

Estimator Random Effects Fixed Effects

E(ui|Xi) = 0 E(ui|Xi) =/= 0
GLS or RE-OLS Consistent and Inconsistent
(Random Effects) Efficient
LSDV or FE-OLS Consistent Consistent
(Fixed Effects) Inefficient Possibly Efficient
Random Effects Model
Hypothesis Testing

 Fixed effects estimator is consistent under H0

and H1; Random effects estimator is efficient
under H0, but it is inconsistent under H1.
 Hausman Test Statistic

   
' 1
H  βˆ RE  βˆ FE Var (βˆ RE )  Var (βˆ FE )  βˆ RE  βˆ FE

~  2 (# βˆ FE ), provided # βˆ FE  # βˆ RE (no intercept )

Random Effects Model
Hypothesis Testing

 Alternative Hausman Test

 Estimate the random effects model
yit  xit' β  xi' γ  (ui  eit )
 F Test that  = 0
H 0 : γ  0  H 0 : Cov(ui , xit )  0
Extensions

 Random Coefficients Model

yit  xit' β i eit 
 it
y  x '
β  ( x it u i eit )
'

βi  β  u i 
it

 Mixed Effects Model

yit  xit' β  (z it' γ i eit )

 Two-Way Effects yit  xit' β  u i vt  eit

 Nested Random Effects yijt  xijt β  u i vij  eijt

'
Example: U. S. Productivity
 Munnell [1988] Productivity Data
48 Continental U.S. States, 17 Years:1970-
1986
 STATE = State name,
 ST ABB=State abbreviation,
 YR =Year, 1970, . . . ,1986,
 PCAP =Public capital,
 HWY =Highway capital,
 WATER =Water utility capital,
 UTIL =Utility capital,
 PC =Private capital,
 GSP =Gross state product,
 EMP =Employment,
References
 B. H. Baltagi, Econometric Analysis of Panel Data, 4th
ed., John Wiley, New York, 2008.
 W. H. Greene, Econometric Analysis, 7th ed., Chapter
11: Models for Panel Data, Prentice Hall, 2011.
 C. Hsiao, Analysis of Panel Data, 2nd ed., Cambridge
University Press, 2003.
 J. M. Wooldridge, Econometric Analysis of Cross Section
and Panel Data, The MIT Press, 2002.