Econometric Analysis of Panel Data: William Greene Department of Economics Stern School of Business

Part 20: Selection [1/66]
Econometric Analysis of Panel Data

William Greene
Department of Economics
Stern School of Business
Econometric Analysis of Panel Data
20. Sample Selection and Attrition
Received Sunday, April 27, 2014

I have a paper regarding strategic alliances between firms, and their impact on firm risk. While
observing how a firms strategic alliance formation impacts its risk, I need to correct for two types of
selection biases. The reviews at Journal of Marketing asked us to correct for the propensity of firms to
enter into alliances, and also the propensity to select a specific partner, before we examine how the
partnership itself impacts risk.
Our approach involved conducting a probit of alliance formation propensity, take the inverse mills and
include it in the second selection equation which is also a probit of partner selection. Then, we include
inverse mills from the second selection into the main model. The review team states that this is not
correct, and we need an MLE estimation in order to correctly model the set of three equations. The
Associate Editors point is given below. Can you please provide any guidance on whether this is a valid
criticism of our approach. Is there a procedure in LIMDEP that can handle this set of three equations
with two selection probit models?
AEs comment:
Please note that the procedure of using an inverse mills ratio is only consistent when the main
equation where the ratio is being used is linear. In non-linear cases (like the second probit used by the
authors), this is not correct. Please see any standard econometric treatment like Greene or Wooldridge.
A MLE estimator is needed which will be far from trivial to specify and estimate given error
correlations between all three equations.
Hello Dr. Greene,

My name is xxxxxxxxxx and I go to the University of xxxxxxxx.
I see that you have an errata page on your website of your econometrics book 7th edition.
It seems like you want to correct all mistakes so I think I have spotted a possible
proofreading error.
On page 477 (theorem 13.2) you want to show that theta is consistent and you say that
"But, at the true parameter values, qn(0) 0. So, if (13-7) is true, then it must follow
that qn(GMM) 0 as well because of the identification assumption"
I think in the second line it should be qn(GMM) 0, not 0.
I also have a questions about nonlinear GMM - which is more or less nonlinear IV technique
I suppose.
I am running a panel non-linear regression (non-linear in the parameters) and I have L
parameters and K exogenous variables with L>K.
In particular my model looks kind of like this: Y = b1*X^b2 + e, and so I am trying to
estimate the extra b2 that don't usually appear in a regression.
From what I am reading, to run nonlinear GMM I can use the K exogenous variables to
construct the orthogonality conditions but what should I use for the extra, b 2 coefficients?
Just some more possible IVs (like lags) of the exogenous variables?
I agree that by adding more IVs you will get a more efficient estimation, but isn't it only the
case when you believe the IVs are truly uncorrelated with the error term?
So by adding more "instruments" you are more or less imposing more and more restrictive
assumptions about the model (which might not actually be true).
I am asking because I have not found sources comparing nonlinear GMM/IV to nonlinear
least squares. If there is no homoscadesticity/serial correlation what is more efficient/give
tighter estimates?
Dueling Selection Biases

From two emails, same day.
I am trying to find methods which can deal

with data that is non-randomised and
suffers from selection bias.
bias
I explain the probability of answering
questions using, amongother independent
variables, a variable which measures
knowledge breadth.Knowledge
breadthcan be constructed only for those
individuals that fill in a skill description in
the company intranet. This is where
theselection bias comes from.
The Crucial Element
Selection on the unobservables
Selection into the sample is based on both

observables and unobservables
All the observables are accounted for
Unobservables in the selection rule also appear
in the model of interest (or are correlated with
unobservables in the model of interest)
Selection Bias=the bias due to not

accounting for the unobservables that link
the equations.
A Sample Selection Model
Linear model
2 step
ML Murphy & Topel
Binary choice application
Other models
Canonical Sample Selection Model

Regression Equation
y*=x+
Sample Selection Mechanism
d*=z+u; d=1[d* > 0] (probit)
y = y* if d = 1; not observed otherwise
Is the sample 'nonrandomly selected?'
E[y*|x,d=1] = x+E[ | x, d 1]
= x+E[ | x,u z]
= x something if Cor[,u|x] 0
A left out variable problem (again)
Incidental truncation
Applications
Labor Supply model:
y*=wage-reservation wage
d=labor force participation
Attrition model: Clinical studies of medicines

Survival bias in financial data
Income studies value of a college application
Treatment effects
Any survey data in which respondents self select
to report
Etc
Estimation of the Selection Model
Two step least squares
Inefficient
Simple exists in current software
Simple to understand and widely used
Full information maximum likelihood
Efficient
Simple exists in current software
Not so simple to understand widely
misunderstood
Heckmans Model
i +i
yi *=x
i +ui ; d=1[d*
d*=
z
> 0] (probit)
i
i
i
yi = yi * if di = 1; not observed otherwise
[i ,ui ]~Bivariate Normal[0,0,2 , ,1]
i +E[i | xi , di 1]
E[yi *|xi ,d=1]
= x
i
i +E[i | xi ,ui z i ]
= x
(z
i )
(
z
)
i
= x
i i
= x
Least squares is biased and inconsistent again. Left out variable
Two Step Estimation

Step 1: Estimate the probit model
i +ui ; d=1[d*
d*=
z
> 0] (probit).
i
i
i
(z
i )
. Now compute i
Estimation of by
i )
(z
Step 2: Estimate the regression model with estimated regressor
i +i
yi *=x
i +E[i | xi , di 1]
E[yi *|xi ,d=1]
= x
i
The LAMBDA
i i
= x
i.
Linearly regress yi on xi ,
Step2a. Fix standard errors (Murphy and Topel). Estimate
and using and e'e/n
FIML Estimation
logL d0 log zi
z /
i2
i
d1 log
exp
i
2
2
2
1
2

i
i yi x
1
Let
1 / , =- / , =
1-2
logL d0 log zi
2
1
i ( 1 2 )z i (yi x
exp yi x
)
i
2
2
Note : no inverse Mills ratio appears anywhere in the model.

d1 log
Classic Application
Mroz, T., Married womens labor supply,

Econometrica, 1987.
A (my) specification
N =753
N1 = 428
LFP=f(age,age2,family income, education,
kids)
Wage=g(experience, exp2, education, city)
Two step and FIML estimation
Selection Equation
+---------------------------------------------+
| Binomial Probit Model
|
| Dependent variable
LFP
|
| Number of observations
753
|
| Log likelihood function
-490.8478
|
+---------------------------------------------+
+--------+--------------+----------------+--------+--------+----------+
|Variable| Coefficient | Standard Error |b/St.Er.|P[|Z|>z]| Mean of X|
+--------+--------------+----------------+--------+--------+----------+
---------+Index function for probability
Constant|
-4.15680692
1.40208596
-2.965
.0030
AGE
|
.18539510
.06596666
2.810
.0049
42.5378486
AGESQ
|
-.00242590
.00077354
-3.136
.0017
1874.54847
FAMINC |
.458045D-05
.420642D-05
1.089
.2762
23080.5950
WE
|
.09818228
.02298412
4.272
.0000
12.2868526
KIDS
|
-.44898674
.13091150
-3.430
.0006
.69588313
Heckman Estimator and MLE
Extension Treatment Effect

What is the value of an elite college education?
i +ui ; d=1[d*
d*=
z
> 0] (probit)
i
i
i
i +di i observed for everyone
yi *=x
[i ,ui ]~Bivariate Normal[0,0,2 , ,1]
i +d+E[
E[yi *|xi ,d=1]
= x
i | xi , di 1]
i
i
i ++E[i | xi ,ui z i ]
= x
(z
i )
i )
(z
i +
= x
i + i
= x
i +d+E[
E[yi *|xi ,d=0]
= x
i | xi , di 0]
i
i
(zi )
i )
(z
i
= x
Least squares is still biased and inconsistent. Left out variable
Sample Selection
An approach modeled on Heckman's model
Regression Equation:
Prob[y=j|x,u]=P(); =exp(x+u)
Selection Equation:
d=1[z+>0]
(The usual probit)
[u,]~n[0,0,1,1,] (Var[u] is absorbed in )
Estimation:
Nonlinear Least Squares: [Terza (1998, see cite in text).]
2 (z+)
E[y|x,d=1]=exp(x+
)
(z)
FIML using Hermite quadrature: [Greene (Stern wp, 97-02, 1997)]
Extensions Binary Data

Application: In German Health care data, insurance choices
i +ui ; d=1[d*
d*=
z
> 0] (probit)
i
i
i
i +i , yi=1[yi * > 0] (probit)
yi *=x
[i ,ui ]~Bivariate Normal[0,0,1, ,1]
Estimation:
(1) Two step? (Wooldridge text)
(2) FIML (Stata, LIMDEP)
Panel Data and Selection

Selection equation with time invariant individual effect
it i it 0]
dit 1[z
Observation mechanism: (yit , xit ) observed when dit 1
Primary equation of interest
Common effects linear regression model
it i it
yit | (dit 1) x
"Selectivity" as usual arises as a problem when the unobservables
are correlated; Corr(it , it ) 0.
The common effects, i and i make matters worse.
Panel Data and Sample Selection Models:

A Nonlinear Time Series
I. 1990-1992: Fixed and Random Effects
Extensions
II. 1995 and 2005: Model Identification through
Conditional Mean Assumptions
III. 1997-2005: Semiparametric Approaches based
on Differences and Kernel Weights
IV. 2007: Return to Conventional Estimators, with
Bias Corrections
Panel Data Sample Selection Models

Verbeek, Economics Letters, 1990.
it wi it 0](Random effects probit)
dit 1[z
it i it ; (Fixed effects regression)
yit | (dit 1) x
Proposed "marginal likelihood" based on joint normality
it
z+
it ui,1 ditui,2
f(ui,1 ,ui,2 )dui,1dui,2
logLi t1 (2dit 1)

2
2
(1 dit )
it ( / )dit (yit yi ) ( xit xi )'

(Integrate out the random effects; difference out the fixed effects.)
Ti
ui,1 ,ui,2 are time invariant uncorrelated standard normal variables

How to do the integration? Natural candidate for simulation.
(Not mentioned in the paper. Too early.)
[Verbeek and Nijman: Selectivity "test" based on this model, International
Economic Review, 1992.]
Zabel Economics Letters
Inappropriate to have a mix of FE and RE

models
Two part solution
Treat both effects as fixed

Project both effects onto the group means of
the variables in the equations
Resulting model is two random effects
equations
Use both random effects
Selection with Fixed Effects

yit * i xit it , i xi wi , wi ~ N [0,1]
dit * i zit uit , i zi vi , vi ~ N [0,1]
(it , uit ) ~ N 2 [(0, 0), ( 2 ,1, )].
Li
dit 0
zit zi vi (vi )dvi
z z v ( / ) 1
i
i
it
d 1 it
it 2 (vi , wi )dvi dwi
-
it
1 2

it yit xit xi wi
Practical Complications
The bivariate normal integration is actually the
product of two univariate normals, because in the
specification above, vi and wi are assumed to be
uncorrelated. Vella notes, however, given the
computational demands of estimating by maximum
likelihood induced by the requirement to evaluate
multiple integrals, we consider the applicability of
available simple, or two step procedures.
Simulation
The first line in the log likelihood is of the form
Ev[d=0()] and the second line is of the form
Ew[Ev[()()/]]. Using simulation instead, the
simulated likelihood is
LSi
1 R
R r 1
dit 0
zit zi vi ,r
zit zi vi ,r ( / )it ,r 1 it ,r
dit 1
yit xit xi wi ,r
1 R
R r 1
it ,r
Correlated Effects
Suppose that wi and vi are bivariate standard normal with
correlation vw. We can project wi on vi and write
wi = vwvi + (1-vw2)1/2hi
where hi has a standard normal distribution. To allow
the correlation, we now simply substitute this expression
for wi in the simulated (or original) log likelihood, and
add vw to the list of parameters to be estimated. The
simulation is then over still independent normal variates,
vi and hi.
Conditional Means
A Feasible Estimator
Estimation
Kyriazidou - Semiparametrics
Assume 2 periods
Estimate selection equation by FE logit
Use first differences and weighted least squares:
= N d d x x 1 N d d
i xi y i
i=1
i1
i2
i
i
i
i=1
i1
i2
1 w i
i K
kernel function.
h h
Use with longer panels - any pairwise differences
Extensions based on pairwise differences by RochinaBarrachina and Dustman/Rochina-Barrachina (1999)
Bias Corrections
Val and Vella, 2007 (Working paper)

Assume fixed effects
Bias corrected probit estimator at the first

step
Use fixed probit model to set up second step
Heckman style regression treatment.
Postscript
What selection process is at work?
All of the work examined here (and in the literature)

assumes the selection operates anew in each period
An alternative scenario: Selection into the panel, once, at
baseline.
Why arent the time invariant components

correlated? (Greene, 2007, NLOGIT development)
Other models
All of the work on panel data selection assumes the main

equation is a linear model.
Any others? Discrete choice? Counts?
Sample Selection
TECHNICAL EFFICIENCY ANALYSIS CORRECTING FOR

BIASES FROM OBSERVED AND UNOBSERVED
VARIABLES: AN APPLICATION TO A NATURAL RESOURCE
MANAGEMENT PROJECT
Empirical Economics: Volume 43, Issue 1 (2012), Pages 55-72
Boris Bravo-Ureta
University of Connecticut
Daniel Solis
University of Miami
William Greene
New York University
The MARENA Program in Honduras

Several programs have been implemented to address
resource degradation while also seeking to improve
productivity, managerial performance and reduce
poverty (and in some cases make up for lack of public
support).
One such effort is the Programa Multifase de Manejo de
Recursos Naturales en Cuencas Prioritarias or
MARENA
in Honduras focusing on small scale hillside
farmers.
OVERALL CONCEPTUAL FRAMEWORK

MARENA
Training &
Financing
Natural, Human &
Social Capital
Off-Farm
Income
More Production
and Productivity
More Farm
Income
Sustainability
Working HYPOTHESIS: if farmers receive private

benefits (higher income) from project activities (e.g.,
training, financing) then adoption is likely to be
sustainable and to generate positive externalities.
The MARENA Program

COMPONENT I: Strengthening Strategic Management Capabilities among
Govt. Institutions (central and local)
COMPONENT II: Support to Nat. Res. Management. Projects
Module 1: Promotion and Organization
Modulo 2: Strengthening Local Institutions & Organizations
Module 3: Investment (farm, municipal & regional)
COMPONENT III: Administration and Supervision
Component II - Module 3
Component II - Module 3 focused on promoting investments in sustainable production
systems with a budget of US $7.6 million (Bravo-Ureta, 2009).
The major activities undertaken with beneficiaries: training in business
management and sustainable farming practices; and the provision of funds to
co-finance investment activities through local rural savings associations (cajas rurales).
Conclusions
Rural poverty in Honduras, largely due to policy

driven unsustainable land use => environmental
degradation, productivity losses, food insecurity,
growing climatic vulnerability (GEF-IFAD, 2002).
Expected Impact Evaluation
Methods
A matched group of beneficiaries and control
farmers is determined using Propensity Score
Matching techniques to mitigate biases that
would stem from selection on observed
variables.
In addition, we deal with possible self-selection
on unobservables arising from unobserved
variables using a selectivity correction model for
stochastic frontiers introduced by Greene (2010).
A Sample Selected SF Model

di = 1[zi + hi > 0], hi ~ N[0,12]
yi = + xi + i, i ~ N[0,2]
(yi,xi) observed only when di = 1.
i = vi - u i
ui = u|Ui| where Ui ~ N[0,12]
vi = vVi where Vi ~ N[0,12].
(hi,vi) ~ N2[(0,1), (1, v, v2)]
Simulated logL for the Standard SF Model

exp[ 12 ( yi x i u |U i |) 2 / v2 ]
f ( yi | x i ,| U i |)
v 2
f ( yi | x i )
|U i |
exp[ 12 ( yi x i u |U i |)2 / v2 ]
p(| U i |)d | U i |
v 2
2exp[ 12 | U i |2 ]
p (| U i |)
, |U i | 0. (Half normal)
2
1 R exp[ 12 ( yi x i u |U ir |) 2 / v2 ]
f ( y | xi )
R r 1
v 2
1 R exp[ 12 ( yi x i u |U ir |) 2 / v2 ]
logLS (,,u ,v ) = i =1 log r 1
v 2
R
This is simply a linear regression with a random constant

term, i = - u |Ui |
Likelihood For a Sample Selected SF Model

f yi | ( x i , d i , z i ,| U i |)
exp 12 ( yi x i u | U i |) 2 / v2 )
v 2
di
( y x | U |) / z
i
i
u
i
(1 d i ) ( zi )
f yi | ( x i , d i , z i )
|U i |
f yi | ( x i , d i , zi ,| U i |) f (| U i |)d | U i |
Simulated Log Likelihood for a Selectivity

Corrected Stochastic Frontier Model
The simulation is over the inefficiency term.
log LS (, , u , v , , ) i 1 log
N
1 R
R r 1
exp 12 ( yi x i u | U ir |) 2 / v2 )
di
v 2
( y x | U |) / z
i
i
u
ir
(1 d i ) ( zi )
JLMS Estimator of ui
exp 12 ( yi x i u | U ir |) 2 / v2 )
fir
v 2
( y x | U |) / a
i
i
u
ir
v
i
1

A = 1 R ( | U |) f , B 1 R f
i
u
ir
ir
i
ir
R r 1
R r 1
Ai
ui Estimator of E[ui |i ]
Bi
R
fir
r 1 g ir | uU ir | where g ir R
,
r 1 fir
R
r 1
g ir 1
Closed Form for the Selection Model

The selection model can be estimated without
simulation
The stochastic frontier model with correction
for sample selection revisited. Lai, Hung-pin.
Forthcoming, JPA
Based on closed skew normal distribution
Similar to Maddalas 1982 result for the linear
selection model. See slide 42.
Not more computationally efficient.
Statistical properties identical.
Suggested possibility that simulation chatter is an element of
inefficiency in the maximum simulated likelihood estimator.
Closed Form vs. Simulation

Spanish Dairy Farms: Selection based on being farm #1-125. 6 periods
The theory works.
Variables Used
in the Analysis
Production
Participation
Findings from the First Wave
A Panel Data Model

Selection takes place only at the baseline.
There is no attrition.
d i 0 1[zi 0 hi 0 > 0]
Sample Selector
yit wi x it vit uit , t 0,1,... Stochastic Frontier
Selection effect is exerted on wi ; Corr(hi 0 , wi ,)
P ( yit , d i 0 ) P (d i 0 ) P ( yit | d i 0 )
Conditioned on the selection (hi 0 ) observations are independent.
P ( yi 0 , yi1 ,..., yiT | d i 0 ) t 0 P ( yit | d i 0 )
T
I.e., the selection is acting like a permanent random effect.

P ( yi 0 , yi1 ,..., yiT , d i 0 ) P (d i 0 ) t 0 P( yit | d i 0 )
T
Simulated Log Likelihood

log LS ,C (, , u , v , )
1 R
d 1 log r 1
i
R
T
t 0
exp 12 ( yit xit u | U itr |) 2 / v2 )
v 2
( yit xit u | U itr |) / v ai 0
Main Empirical Conclusions from Waves 0 and 1
Benefit group is more efficient in both years

The gap is wider in the second year
Both means increase from year 0 to year 1
Both variances decline from year 0 to year 1
Attrition
In a panel, t=1,,T individual I leaves the

sample at time Ki and does not return.
If the determinants of attrition (especially

the unobservables) are correlated with
the variables in the equation of interest,
then the now familiar problem of sample
selection arises.
Application of a Two Period Model
Hemoglobin and Quality of Life in Cancer

Patients with Anemia,
Finkelstein (MIT), Berndt (MIT), Greene
(NYU), Cremieux (Univ. of Quebec)
1998
With Ortho Biotech seeking to change
labeling of already approved drug
erythropoetin.
r-HuEPO
QOL Study
Quality of life study
yit = self administered quality of life survey, scale = 0,,100

xit = hemoglobin level, other covariates
Treatment effects model (hemoglobin level)

Background r-HuEPO treatment to affect Hg level
Important statistical issues
i = 1, 1200+ clinically anemic cancer patients undergoing

chemotherapy, treated with transfusions and/or r-HuEPO
t = 0 at baseline, 1 at exit. (interperiod survey by some patients
was not used)
Unobservable individual effects

The placebo effect
Attrition sample selection
FDA mistrust of community based not clinical trial based
statistical evidence
Objective when to administer treatment for maximum

marginal benefit
Dealing with Attrition
The attrition issue: Appearance for the second

interview was low for people with initial low QOL
(death or depression) or with initial high QOL
(dont need the treatment). Thus, missing data at
exit were clearly related to values of the
dependent variable.
Solutions to the attrition problem
Heckman selection model (used in the study)
Prob[Present at exit|covariates] = (z) (Probit model)

Additional variable added to difference model i =
(zi)/(zi)
The FDA solution: fill with zeros. (!)
An Early Attrition Model

Hausman, J . and Wise, D., "Attrition Bias in Experimental and Panel Data:
The Gary Income Maintenance Experiment," Econometrica, 1979.
A two period model:
Structural response model (Random Effects Regression)
i1 i1 ui
yi1 x
i2 i2 ui
yi2 x
Attrition model for observation in the second period (Probit)
i2 w i2 vi2
zi2* yi2 x
zi2 1(zi2* 0)
Endogeneity "problem"
12 Corr[i1 ui , i2 ui ] u2 /(2 u2 )
Corr[vi2 , i2 ui ]
Corr[vi2 (i2 ui ), i2 ui )
Methods of Estimating the Attrition

Model
Heckman style selection model

Two step maximum likelihood
Full information maximum likelihood
Two step method of moments estimators
Weighting schemes that account for the
survivor bias
Selection Model
Reduced form probit model for second period observation equation
zi2* xi2 ( ) wi (i2 ui vi )
ri2 hi2
zi2
1(zi2* 0)
Conditional means for observations observed in the second period

(ri2 )
E[yi2 | xi2 , zi2 1] xi2 (12 )
(ri2 )
First period conditional means for observations observed in the
second period
E[yi1 | xi1 , zi2 1] xi1 (12 )
(ri2 )
(ri2 )
(1) Estimate probit equation

(2) Combine these two equations with a period dummy variable, use
OLS with a constructed regressor in the second period
THE TWO DI STURBANCES ARE CORRELATED.
TREAT THI S I S A SUR MODEL. (EQUI VALENT TO MDE)
Maximum Likelihood
(yi1 xi1)2
log2
log
2
22
LogLi
zi2
[(yi2 12yi1) (x12 12xi1))]2

log log 1
2
22 (1 12
)
r ( / )(y x )
i2
i2
log i2
2
2
12
ri2 (12 / )(yi1 xi1)
(1 zi2 ) log
2
1 12
(1) See H&W for FIML estimation
(2) Use the invariance principle to reparameterize

(3) Estimate separately and use a two step ML with Murphy and
Topel correction of asymptotic covariance matrix.
A Model of Attrition
Nijman and Verbeek, Journal of Applied

Econometrics, 1992
Consumption survey (Holland, 1984
1986)
Exogenous selection for participation (rotating

panel)
Voluntary participation (missing not at
random attrition)
Attrition Model
The main equation
yi,t 0 xi,t %i i,t , Random effects consumption function
%i xi ui ,
Mundlak device; ui uncorrelated with X i
yi,t 0 xi,t xi ui i,t , Reduced form random effects model
The selection mechanism
ait 1[individual i asked to participate in period t] Purely exogenous
ait may depend on observables, but does not depend on unobservables
rit 1[individual i chooses to participate if asked] Endogenous.
rit is the endogenous participation dummy variable
ait 0 rit 0
ait 1 the selection mechanism operates
Selection Equation
The main equation
yi,t 0 xi,t xi ui i,t , Reduced form random effects model
The selection mechanism
rit 1[individual i chooses to participate if asked] Endogenous.
rit is the endogenous participation dummy variable
ait 0 rit 0
ait 1 the selection mechanism operates
rit 1[ 0 xi,t xi zi,t v i w i,t 0] all observed if ait 1
State dependence: z may include ri,t-1
Latent persistent unobserved heterogeneity: 2v 0.
"Selection" arises if Cov[i,t ,w i,t ] 0 or Cov[ui,v i ] 0
Estimation Using One Wave
Use any single wave as a cross section

with observed lagged values.
Advantage: Familiar sample selection
model
Disadvantages
Loss of efficiency
One can no longer distinguish between state
dependence and unobserved heterogeneity.
One Wave Model

A standard sample selection model.
y it 0 xit xi (ui it )
rit 1[ 0 xit xi 1ri,t 1 2ai,t 1 (v i w it ) 0]
With only one period of data and ri,t-1 exogenous,
this is the Heckman sample selection mod el.
If > 0, then ri,t-1 is correlated with v i and the Heckman
approach fails.
An assumption is required:
(1) Include ri,t-1 and assume no unobserved heterogeneity
(2) Exclude ri,t-1 and assume there is no state dependence.
In either case, now if Cov[(ui it ),(v i w it )] we can use OLS.
Otherwise, use the maximum likelihood estimator.
Maximum Likelihood Estimation
See Zabels model in slides 20 and 23.

Because numerical integration is required in
one or two dimensions for every individual in
the sample at each iteration of a high
dimensional numerical optimization problem,
this is, though feasible, not computationally
attractive.
The dimensionality of the optimization is irrelevant

This is much easier in 2015 than it was in 1992
(especially with simulation) The authors did the
computations with Hermite quadrature.
Testing for Selection?
Maximum Likelihood Results
Covariances were highly insignificant.

LR statistic=0.46.
Two step results produced the same

conclusion based on a Hausman test
ML Estimation results looked like the two
step results.
A Dynamic Ordered Probit Model
Random Effects Dynamic

Ordered Probit Model
Random Effects Dynamic Ordered Probit Model
hit * xit Jj1 jhi,t 1( j) i i,t
hi,t j if j-1 < hit * < j
hi,t ( j) 1 if hi,t = j
Pit,j P[hit j] ( j xit Jj1 jhi,t 1( j) i )
( j1 xit Jj1 jhi,t 1( j) i )
Parameterize Random Effects
i 0 Jj11,jhi,1( j) xi ui
Simulation or Quadrature Based Estimation
lnL= i=1 ln
N
Ti
t 1
Pit,j f( j )d j
A Study of Health Status in the Presence

of Attrition
THE DYNAMICS OF HEALTH IN THE BRITISH
HOUSEHOLD PANEL SURVEY,
Contoyannis, P., Jones, A., N. Rice
Journal of Applied Econometrics, 19, 2004,
pp. 473-503.
Self assessed health
British Household Panel Survey (BHPS)
1991 1998 = 8 waves

About 5,000 households
Attrition
Testing for Attrition Bias
Three dummy variables added to full model with unbalanced

panel suggest presence of attrition effects.
Attrition Model with IP Weights
Assumes (1) Prob(attrition|all data) = Prob(attrition|selected variables) (ignorability)

(2) Attrition is an absorbing state. No reentry.
Obviously not true for the GSOEP data above.
Can deal with point (2) by isolating a subsample of those present at wave 1 and the
monotonically shrinking subsample as the waves progress.
Probability Weighting Estimators
A Patch for Attrition

(1) Fit a participation probit equation for
each wave.
(2) Compute p(i,t) = predictions of
participation for each individual in each
period.
Special assumptions needed to make this work
Ignore common effects and fit a weighted

pooled log likelihood: i t [dit/p(i,t)]logLPit.
Inverse Probability Weighting

Panel is based on those present at WAVE 1, N1 individuals
Attrition is an absorbing state. No reentry, so N1 N2 ... N8.
Sample is restricted at each wave to individuals who were present at
the previous wave.
d it = 1[Individual is present at wave t].
d i1 = 1 i, d it 0 d i ,t 1 0.
x%
i1 covariates observed for all i at entry that relate to likelihood of
being present at subsequent waves.
(health problems, disability, psychological well being, self employment,
unemployment, maternity leave, student, caring for family member, ...)
Probit model for d it 1[x%

i1 wit ], t = 2,...,8. it fitted probability.
t
Assuming attrition decisions are independent, Pit s 1 is
d it
Inverse probability weight W
it
Pit
Weighted log likelihood
logLW i 1 t 1 log Lit (No common effects.)

N
Spatial Autocorrelation in a
Sample Selection Model
Flores-Lagunes, A. and Schnier, K., Sample selection and Spatial Dependence, Journal of Applied
Econometrics, 27, 2, 2012, pp. 173-204.
Alaska Department of Fish and Game.

Pacific cod fishing eastern Bering Sea grid of locations
Observation = catch per unit effort in grid square
Data reported only if 4+ similar vessels fish in the region
1997 sample = 320 observations with 207 reported full data
Flores-Lagunes, A. and Schnier, K., Sample selection and Spatial Dependence, Journal of Applied
Econometrics, 27, 2, 2012, pp. 173-204.
LHS is catch per unit effort = CPUE

Site characteristics: MaxDepth, MinDepth, Biomass
Fleet characteristics:
Catcher vessel (CV = 0/1)

Hook and line (HAL = 0/1)
Nonpelagic trawl gear (NPT = 0/1)
Large (at least 125 feet) (Large = 0/1)
yi*1 0 xi1 ui1
ui1 j i cij u j1 i1
yi*2 0 xi 2 ui 2
ui 2 j i cij u j 2 i 2
i1
~N
i 2
0

0
12
,
12
12
, (?? 1 1??)
2
2
Observation Mechanism
yi1 1 yi*1 > 0
Probit Model
yi 2 yi*2 if yi1 = 1, unobserved otherwise.
u1 Cu1 1
C = Spatial weight matrix, Cii 0.
u1 [I C]1 1 = (1) 1 , likewise for u 2
()
, Var[u ] ( )
Cov[u , u ] () ( )
y 0 xi1 j 1 () i1 , Var[ui1 ]
*
i1
(1)
ij
y 0 xi 2 j 1 ( )ij(2)
*
i2
2
1
i2
i2
j 1
2
1
j 1
N
i1
i2
12
j 1
(1) 2
ij
( 2) 2
ij
(1)
ij
(2)
ij
Spatial Weights
1
cij 2 ,
dij
dij Euclidean distance
Band of 7 neighbors is used
Row standardized.
Two Step Estimation
Probit estimated by Pinske/Slade GMM
0 xi1
() ( )
()
N
(1)
ij
j 1
j 1
(2)
ij
(1) 2
ij
2
1
()
N
j 1
0 xi1
(1) 2
ij
j 1 ()
Spatial regression with included IMR in second step

(*) GMM procedure combines the two steps in one large estimation.
2
1
(1) 2
ij

Econometric Analysis of Panel Data: William Greene Department of Economics Stern School of Business

Uploaded by

Copyright:

Available Formats

Econometric Analysis of Panel Data: William Greene Department of Economics Stern School of Business

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Econometric Analysis of Panel Data: William Greene Department of Economics Stern School of Business

Uploaded by

Copyright:

Available Formats

Part 20: Selection [1/66]

Econometric Analysis of Panel Data

Part 20: Selection [2/66]

Econometric Analysis of Panel Data

20. Sample Selection and Attrition

Part 20: Selection [3/66]

Received Sunday, April 27, 2014

Part 20: Selection [4/66]

Hello Dr. Greene,

Part 20: Selection [5/66]

Part 20: Selection [6/66]

Part 20: Selection [7/66]

Dueling Selection Biases

I am trying to find methods which can deal

Part 20: Selection [8/66]

The Crucial Element

Selection on the unobservables

Selection into the sample is based on both

Selection Bias=the bias due to not

Part 20: Selection [9/66]

A Sample Selection Model

Part 20: Selection [10/66]

Canonical Sample Selection Model

Part 20: Selection [11/66]

Labor Supply model:

Attrition model: Clinical studies of medicines

Part 20: Selection [12/66]

Estimation of the Selection Model

Two step least squares

Full information maximum likelihood

Part 20: Selection [13/66]

Part 20: Selection [14/66]

Two Step Estimation

Part 20: Selection [15/66]

Note : no inverse Mills ratio appears anywhere in the model.

Part 20: Selection [16/66]

Mroz, T., Married womens labor supply,

Two step and FIML estimation

Part 20: Selection [17/66]

Part 20: Selection [18/66]

Heckman Estimator and MLE

Part 20: Selection [19/66]

Extension Treatment Effect

Least squares is still biased and inconsistent. Left out variable

Part 20: Selection [20/66]

Part 20: Selection [21/66]

Extensions Binary Data

Part 20: Selection [22/66]

Panel Data and Selection

Part 20: Selection [23/66]

Panel Data and Sample Selection Models:

Part 20: Selection [24/66]

Panel Data Sample Selection Models

it ( / )dit (yit yi ) ( xit xi )'

ui,1 ,ui,2 are time invariant uncorrelated standard normal variables

Part 20: Selection [25/66]

Zabel Economics Letters

Inappropriate to have a mix of FE and RE