Discrete Choice Modeling
William Greene
Stern School of Business
New York University
Part 4
Panel Data Models
Application: Health Care Panel Data
German Health Care Usage Data, 7,293 Individuals, Varying Numbers of Periods
Variables in the file are
Data downloaded from Journal of Applied Econometrics Archive. This is an unbalanced panel with 7,293
individuals. They can be used for regression, count models, binary choice, ordered choice, and bivariate binary
choice. This is a large data set. There are altogether 27,326 observations. The number of observations
ranges from 1 to 7. (Frequencies are: 1=1525, 2=2158, 3=825, 4=926, 5=1051, 6=1000, 7=987). Note, the
variable NUMOBS below tells how many observations there are for each person. This variable is repeated in each
row of the data for the person. (Downloaded from the JAE Archive)
DOCTOR = 1(Number of doctor visits > 0)
HOSPITAL = 1(Number of hospital visits > 0)
HSAT = health satisfaction, coded 0 (low) - 10 (high)
DOCVIS = number of doctor visits in last three months
HOSPVIS = number of hospital visits in last calendar year
PUBLIC = insured in public health insurance = 1; otherwise = 0
ADDON = insured by add-on insurance = 1; otherswise = 0
HHNINC = household nominal monthly net income in German marks / 10000.
(4 observations with income=0 were dropped)
HHKIDS = children under age 16 in the household = 1; otherwise = 0
EDUC = years of schooling
AGE = age in years
MARRIED = marital status
EDUC = years of education
Unbalanced Panel: Group Sizes
Panel Data Models
Benefits
Modeling heterogeneity
Rich specifications
Modeling dynamic effects in individual behavior
Costs
More complex models and estimation procedures
Statistical issues for identification and estimation
Fixed and Random Effects
Model: Feature of interest y
it
Probability distribution or conditional mean
Observable covariates x
it
, z
i
Individual specific heterogeneity, u
i
Probability or mean, f(x
it
,z
i
,u
i
)
Random effects: E[u
i
|x
i1
,,x
iT
,z
i
] = 0
Fixed effects: E[u
i
|x
i1
,,x
iT
,z
i
] = g(X
i
,z
i
).
The difference relates to how u
i
relates to the
observable covariates.
Household Income
We begin by analyzing Income using linear regression.
Fixed and Random Effects in Regression
y
it
= a
i
+ bx
it
+ e
it
Random effects: Two step FGLS. First step is OLS
Fixed effects: OLS based on group mean differences
How do we proceed for a binary choice model?
y
it
* = a
i
+ bx
it
+ e
it
y
it
= 1 if y
it
* > 0, 0 otherwise.
Neither ols nor two step FGLS works (even
approximately) if the model is nonlinear.
Models are fit by maximum likelihood, not OLS or GLS
New complications arise that are absent in the linear case.
Pooled Linear Regression - Income
----------------------------------------------------------------------
Ordinary least squares regression ............
LHS=HHNINC Mean = .35208
Standard deviation = .17691
Number of observs. = 27326
Model size Parameters = 2
Degrees of freedom = 27324
Residuals Sum of squares = 796.31864
Standard error of e = .17071
Fit R-squared = .06883
Adjusted R-squared = .06879
Model test F[ 1, 27324] (prob) = 2019.6(.0000)
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
Constant| .12609*** .00513 24.561 .0000
EDUC| .01996*** .00044 44.940 .0000 11.3206
--------+-------------------------------------------------------------
Fixed Effects
----------------------------------------------------------------------
Least Squares with Group Dummy Variables..........
Ordinary least squares regression ............
LHS=HHNINC Mean = .35208
Standard deviation = .17691
Number of observs. = 27326
Model size Parameters = 7294
Degrees of freedom = 20032
Residuals Sum of squares = 277.15841
Standard error of e = .11763
Fit R-squared = .67591
Adjusted R-squared = .55791
Model test F[***, 20032] (prob) = 5.7(.0000)
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
EDUC| .03664*** .00289 12.688 .0000 11.3206
--------+-------------------------------------------------------------
For the pooled model, R squared was .06883 and the estimated coefficient
On EDUC was .01996.
Random Effects
----------------------------------------------------------------------
Random Effects Model: v(i,t) = e(i,t) + u(i)
Estimates: Var[e] = .013836
Var[u] = .015308
Corr[v(i,t),v(i,s)] = .525254
Lagrange Multiplier Test vs. Model (3) =*******
( 1 degrees of freedom, prob. value = .000000)
(High values of LM favor FEM/REM over CR model)
Baltagi-Li form of LM Statistic = 4534.78
Sum of Squares 796.363710
R-squared .068775
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
EDUC| .02051*** .00069 29.576 .0000 11.3206
Constant| .11973*** .00808 14.820 .0000
--------+-------------------------------------------------------------
Note: ***, **, * = Significance at 1%, 5%, 10% level.
----------------------------------------------------------------------
For the pooled model, the estimated coefficient on EDUC was .01996.
Fixed vs. Random Effects
Linear Models
Fixed Effects
Robust to both cases
Use OLS
Convenient
Random Effects
Inconsistent in FE case:
effects correlated with X
Use FGLS: No necessary
distributional assumption
Smaller number of parameters
Inconvenient to compute
Nonlinear Models
Fixed Effects
Usually inconsistent because
of IP problem
Fit by full ML
Extremely inconvenient
Random Effects
Inconsistent in FE case :
effects correlated with X
Use full ML: Distributional
assumption
Smaller number of parameters
Always inconvenient to
compute
Binary Choice Model
Model is Prob(y
it
= 1|x
it
) (z
i
is embedded in x
it
)
In the presence of heterogeneity,
Prob(y
it
= 1|x
it
,u
i
) = F(x
it
,u
i
)
Panel Data Binary Choice Models
Random Utility Model for Binary Choice
U
it
= o + |x
it
+ c
it
+ Person i specific effect
Fixed effects using dummy variables
U
it
= o
i
+ |x
it
+ c
it
Random effects using omitted heterogeneity
U
it
= o + |x
it
+ c
it
+ u
i
Same outcome mechanism: Y
it
= 1[U
it
> 0]
Ignoring Unobserved Heterogeneity
( )
+ c =
'
+ + c
'
= = + c >
' '
= = o =
i it
it i it
it it i it
2
it it u
Assuming strict exogeneity; Cov( ,u ) 0
y *= u
Prob[y 1| x ] Prob[u - ]
Using the same model format:
Prob[y 1| x ] F / 1+ F( )
This is the "population averaged mo
it
it
it
it it
x
x
x
x x
del."
Ignoring Heterogeneity
' '
'
Ignoring heterogeneity, we estimate not .
Partial effects are f( ) not f( )
is underestimated, but f( ) is overestimated.
Which way does it go? Maybe ignoring u is ok?
Not if we want
it it
it
x x
x
to compute probabilities or do
statistical inference about Estimated standard
errors will be too small.
.
Pooled vs. A Panel Estimator
----------------------------------------------------------------------
Binomial Probit Model
Dependent variable DOCTOR
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
Constant| .02159 .05307 .407 .6842
AGE| .01532*** .00071 21.695 .0000 43.5257
EDUC| -.02793*** .00348 -8.023 .0000 11.3206
HHNINC| -.10204** .04544 -2.246 .0247 .35208
--------+-------------------------------------------------------------
Unbalanced panel has 7293 individuals
--------+-------------------------------------------------------------
Constant| -.11819 .09280 -1.273 .2028
AGE| .02232*** .00123 18.145 .0000 43.5257
EDUC| -.03307*** .00627 -5.276 .0000 11.3206
HHNINC| .00660 .06587 .100 .9202 .35208
Rho| .44990*** .01020 44.101 .0000
--------+-------------------------------------------------------------
Partial Effects
----------------------------------------------------------------------
Partial derivatives of E[y] = F[*] with
respect to the vector of characteristics
They are computed at the means of the Xs
Observations used for means are All Obs.
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Elasticity
--------+-------------------------------------------------------------
|Pooled
AGE| .00578*** .00027 21.720 .0000 .39801
EDUC| -.01053*** .00131 -8.024 .0000 -.18870
HHNINC| -.03847** .01713 -2.246 .0247 -.02144
--------+-------------------------------------------------------------
|Based on the panel data estimator
AGE| .00620*** .00034 18.375 .0000 .42181
EDUC| -.00918*** .00174 -5.282 .0000 -.16256
HHNINC| .00183 .01829 .100 .9202 .00101
--------+-------------------------------------------------------------
Effect of Clustering
Y
it
must be correlated with Y
is
across periods
Pooled estimator ignores correlation
Broadly, y
it
= E[y
it
|x
it
] + w
it
,
E[y
it
|x
it
] = Prob(y
it
= 1|x
it
)
w
it
is correlated across periods
Assuming the marginal probability is the same,
the pooled estimator is consistent. (We just saw
that it might not be.)
Ignoring the correlation across periods generally
leads to underestimating standard errors.
Cluster Corrected Covariance Matrix
Robustness is not the justification.
( )( )
( )
c
1
1 1
1 1 1
the number if clusters
n number of observations in cluster c
= inverse of second derivatives matrix
= derivative of log density for observation
1
c c
ic
C n n
ic ic
c i i
C
C
C
= = =
=
=
| |
'
=
|
\ .
H
g
V H g g H
Cluster Correction: Doctor
----------------------------------------------------------------------
Binomial Probit Model
Dependent variable DOCTOR
Log likelihood function -17457.21899
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
| Conventional Standard Errors
Constant| -.25597*** .05481 -4.670 .0000
AGE| .01469*** .00071 20.686 .0000 43.5257
EDUC| -.01523*** .00355 -4.289 .0000 11.3206
HHNINC| -.10914** .04569 -2.389 .0169 .35208
FEMALE| .35209*** .01598 22.027 .0000 .47877
--------+-------------------------------------------------------------
| Corrected Standard Errors
Constant| -.25597*** .07744 -3.305 .0009
AGE| .01469*** .00098 15.065 .0000 43.5257
EDUC| -.01523*** .00504 -3.023 .0025 11.3206
HHNINC| -.10914* .05645 -1.933 .0532 .35208
FEMALE| .35209*** .02290 15.372 .0000 .47877
--------+-------------------------------------------------------------
Modeling a Binary Outcome
Did firm i produce a product or process innovation in year t ?
y
it
: 1=Yes/0=No
Observed N=1270 firms for T=5 years, 1984-1988
Observed covariates: x
it
= Industry, competitive pressures,
size, productivity, etc.
How to model?
Binary outcome
Correlation across time
Heterogeneity across firms
Application 2: Innovation
A Random Effects Model
i
i
it i
i
1 2 ,
1
, u ~ [0, ]
T = observations on individual i
For each period, y 1[ 0] (given u )
Joint probability for T observations is
Prob( , ,...) ( )
For convenience, wr
i
it i u
it
T
i i it it
it
i
t
u N
U
y y F y u
U
=
'
+ o
= >
'
= o+ +
= o+
[
x
x
|
|
i u
1 , u
1
1
ite u = , ~ [0,1]
log | ,... log ( )
It is not possible to maximize log | ,... because of
the unobserved random effects.
i
i i
T N
N it it i
i i t
N
v v N
L v v F y v
L v v
= =
o
(
'
= o+ + o
[
x |
A Computable Log Likelihood
( )
1 1
u
log log ( , )
Maximize this functio
The unobserved heterogeneity is a
n with respect to , , .
How to compute the integral?
(1) Analyticall
verage
y? No, no
d
fo
o t
m
u
u
r
i
T N
it it u i i i
i t
L F y v f v dv
= =
(
'
= o+ + o
o o
[
}
x |
|
la exists.
(2) Approximately, using Gauss-Hermite quadrature
(3) Approximately using Monte Carlo simulation
Quadrature Butler and Moffitt
( )
=
=
=
(
' o + + o |
| |
|
t
\ .
=
[
}
}
| x
i
N
i
i 1
N
i
i
T
it it
1
i
t 1
i
2
u
Th
F(y , v )
g(
v
1 -v
exp
2
2
logL l
is method is used in most commerical software since
og dv
= log dv
(make a change of variable to w = v/ 2
v
198
)
2
=
( )
=
= =
t
~
t
}
N
i
i 1
N H
i 1
h
2
h h
h
h 1
1
l g( 2w)
g( 2z )
The values
og dw
The integral can be
of w (weights) and z
exp -w
(node
computed using
s) are found i
Hermite quadr
n published
tab
ature.
1
les such as A
log w
bramovitz and Stegun (or on the web). H is by
choice. Higher H produces greater accuracy (but takes longer).
Quadrature Log Likelihood
( )
= = =
=
t
(
'
o + + o
[
| x
i
N
h
H
i
T
it it u
1 1 1
h
t h
After all the substitutions, the function to be maximized:
Not simple, but feasibl
1
logL lo F(y , z ) g
e
2
.
w
Simulation
( )
=
=
=
|
| |
(
'
o + + o
|
t
\
=
.
}
[
| x
i
N
i
i 1
i
2
i
T
it it u i
t 1
i
N
i
i 1
N
i
i
i
1
logL log dv
= log dv
T ]
The expected value of the functio
his equals log E[
n of v can be a
F(y ,
ppro
v
-v 1
xima
ex
te
v )
g(v )
d
by dr
g(
aw
p
2
2
v )
ing
= = =
(
'
= o + + o
[
| x
i
T N R
S it it u ir
i
i 1
r
ir
r 1 t 1
R random draws v from the population N[0,1] and
averaging the R functions of v . We maxi
1
logL log F(y , v
i
)
R
mze
Random Effects Model
----------------------------------------------------------------------
Random Effects Binary Probit Model
Dependent variable DOCTOR
Log likelihood function -16290.72192 Random Effects
Restricted log likelihood -17701.08500 Pooled
Chi squared [ 1 d.f.] 2820.72616
Significance level .00000
McFadden Pseudo R-squared .0796766
Estimation based on N = 27326, K = 5
Unbalanced panel has 7293 individuals
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
Constant| -.11819 .09280 -1.273 .2028
AGE| .02232*** .00123 18.145 .0000 43.5257
EDUC| -.03307*** .00627 -5.276 .0000 11.3206
HHNINC| .00660 .06587 .100 .9202 .35208
Rho| .44990*** .01020 44.101 .0000
--------+-------------------------------------------------------------
|Pooled Estimates using the Butler and Moffitt method
Constant| .02159 .05307 .407 .6842
AGE| .01532*** .00071 21.695 .0000 43.5257
EDUC| -.02793*** .00348 -8.023 .0000 11.3206
HHNINC| -.10204** .04544 -2.246 .0247 .35208
--------+-------------------------------------------------------------
Random Parameter Model
----------------------------------------------------------------------
Random Coefficients Probit Model
Dependent variable DOCTOR (Quadrature Based)
Log likelihood function -16296.68110 (-16290.72192)
Restricted log likelihood -17701.08500
Chi squared [ 1 d.f.] 2808.80780
Significance level .00000
McFadden Pseudo R-squared .0793400
Estimation based on N = 27326, K = 5
Unbalanced panel has 7293 individuals
PROBIT (normal) probability model
Simulation based on 50 Halton draws
--------+-------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z]
--------+-------------------------------------------------
|Nonrandom parameters
AGE| .02226*** .00081 27.365 .0000 ( .02232)
EDUC| -.03285*** .00391 -8.407 .0000 (-.03307)
HHNINC| .00673 .05105 .132 .8952 ( .00660)
|Means for random parameters
Constant| -.11873** .05950 -1.995 .0460 (-.11819)
|Scale parameters for dists. of random parameters
Constant| .90453*** .01128 80.180 .0000
--------+-------------------------------------------------------------
Fixed Effects Models
Estimate with dummy variable coefficients
Uit = o
i
+ |xit + c
it
Can be done by brute force for 10,000s of individuals
F(.) = appropriate probability for the observed outcome
Compute | and o
i
for i=1,,N (may be large)
See FixedEffects.pdf in course materials.
1 1
log log ( , )
i
N T
it i it
i t
L F y
= =
'
= o +
x |
Unconditional Estimation
Maximize the whole log likelihood
Difficult! Many (thousands) of parameters.
Feasible NLOGIT (2001) (Brute force)
Fixed Effects Health Model
Groups in which y
it
is always = 0 or always = 1. Cannot compute
i
.
Conditional Estimation
Principle: f(y
i1
,y
i2
, | some statistic) is free of
the fixed effects for some models.
Maximize the conditional log likelihood, given
the statistic.
Can estimate without having to estimate
i
.
Only feasible for the logit model. (Poisson
and a few other continuous variable models.
No other discrete choice models.)
Binary Logit Conditional Probabiities
i
i
1 1 2 2
1 1
1
T
S
1 1
All
Prob( 1| ) .
1
Prob , , ,
exp exp
exp exp
i it
i it
i i
i i
i i
t i
i
t i
it it
i i i i iT iT
T T
it it it it
t t
T T
it it it i
T
it
t
t
d S
t t
e
y
e
Y y Y y Y y
y
d d
y
y
' +
' +
= =
| |
E =
\ .
=
=
=
= =
+
| |
= = =
|
\ .
| | | |
' '
| |
\ . \ .
= =
| | | |
' '
| |
\ . \ .
x
x
x
x x
x x
|
|
|
|
o
o
i
t i
different
i
T
it t=1 i
ways that
can equal S
.
Denominator is summed over all the different combinations of T values
of y that sum to the same sum as the observed . If S is this sum,
there are
it
it
d
y
|
E
E
i
terms. May be a huge number. An algorithm by Krailo
S
and Pike makes it simple.
T
| |
|
\ .
Example: Two Period Binary Logit
i it
i it
i
i
i i
i
t it i
it it
T
it it
T
t 1
i1 i1 i2 i2 iT iT it
T
t 1
it it
d S
t 1
2
i1 i2 it
t 1
i1 i
e
Prob(y 1| ) .
1 e
exp y
Prob Y y , Y y , , Y y y , data .
exp d
Prob Y 0, Y 0 y 0, data 1.
Prob Y 1, Y
' o +
' o +
=
=
E =
=
=
= =
+
| |
'
|
| |
\ .
= = = =
|
|
| |
\ .
'
|
\ .
| |
= = = =
|
\ .
=
x
x
x
x
x
|
|
2
2 it
t 1
2
i1 i2 it
t 1
2
i1 i2 it
t 1
exp( )
0 y 1, data
exp( ) exp( )
exp( )
Prob Y 0, Y 1 y 1, data
exp( ) exp( )
Prob Y 1, Y 1 y 2, data 1.
=
=
=
| | '
= = =
|
' ' +
\ .
| | '
= = = =
|
' ' +
\ .
| |
= = = =
|
\ .
i1
i1 i2
i2
i1 i2
x
x x
x
x x
Comments on Enumeration in the Logit Model
" This can easily be generalized for any T. However
the computations rise geometrically with T and are
excessive for T > 10. See Greene (1993)."
(Baltagi, Panel Data, 1st edition)
"If T is large, getting ... can be cumbersome as
one can guess from 3.5 with T = 3."
(M.J.Lee, "Panel Data Econometrics")
(Both unaware of Krailo and Pike...)
Estimating Partial Effects
The fixed effects logit estimator of | immediately
gives us the effect of each element of x
i
on the log-
odds ratio Unfortunately, we cannot estimate the
partial effects unless we plug in a value for
i
.
Because the distribution of
i
is unrestricted in
particular, E[
i
] is not necessarily zero it is hard to
know what to plug in for
i
. In addition, we cannot
estimate average partial effects, as doing so would
require finding E[(x
it
|+
i
)], a task that apparently
requires specifying a distribution for
i
.
(Wooldridge, 2002)
Binary Logit Estimation
Estimate | by maximizing conditional logL
Estimate o
i
by using the known | in the FOC for the
unconditional logL
Solve for the N constants, one at a time treating | as
known.
No solution when y
it
sums to 0 or T
i
Works if E[o
i
|
i
y
it
] = E[o
i
].
1
exp( )
( ) 0,
1 exp( )
i
T
i it
it it it
t
i it
y P P
=
'
o +|
= =
'
+ o +|
x
x
Logit Constant Terms
' o
= ' o
o
o
= =
+ o
+
i
i
i
i
T
i it
i
t 1
i i i i
Step 1. Estimate with Chamberlain's conditional estimator
Step 2. Treating as if it were known, estimate from the
first order condition
c 1 e e 1
y
T T 1 c
1 e e
it
it
x
x
= =
=
+
= o => o =
' =
o
i i
T T
it
t 1 t 1
t i i it
i i i i
it
i
i
c 1
T c
Estimate 1/ exp( ) log
c exp( ) is treated as known data.
Solve one equation in one unknown for each .
Note there is no solution if y = 0 or 1.
Iterating bac
it
x
k and forth does not maximize logL.
Fixed Effects Logit Health Model:
Conditional vs. Unconditional
Advantages and Disadvantages
of the FE Model
Advantages
Allows correlation of effect and regressors
Fairly straightforward to estimate
Simple to interpret
Disadvantages
Model may not contain time invariant variables
Not necessarily simple to estimate if very large
samples (Stata just creates the thousands of dummy
variables)
The incidental parameters problem: Small T bias
Incidental Parameters Problems:
Conventional Wisdom
General: The unconditional MLE is biased in
samples with fixed T except in special cases
such as linear or Poisson regression (even when
the FEM is the right model).
The conditional estimator (that bypasses
estimation of
i
) is consistent.
Specific: Upward bias (experience with probit
and logit) in estimators of |
What We KNOW - Analytic
Newey and Hahn: MLE converges in probability to a
vector of constants. (Variance diminishes with increase
in N).
Abrevaya and Hsiao: Logit estimator converges to 2|
when T = 2.
Only the case of T=2 for the binary logit model is known
with certainty. All other cases are extrapolations of this
result or speculative.
What We THINK We Know Monte Carlo
Heckman:
Bias in probit estimator is small if T 8
Bias in probit estimator is toward 0 in
some cases
Katz (et al numerous others),
Greene
Bias in probit and logit estimators is large
Upward bias persists even as T 20
Some Familiar Territory A Monte Carlo
Study of the FE Estimator: Probit vs. Logit
Estimates of Coefficients and Marginal
Effects at the Implied Data Means
Results are scaled so the desired quantity being estimated
(|, o, marginal effects) all equal 1.0 in the population.
A Monte Carlo Study of the FE Probit Estimator
Percentage Biases in Estimates of Coefficients, Standard
Errors and Marginal Effects at the Implied Data Means
Bias Correction Estimators
Motivation: Undo the incidental parameters bias in the
fixed effects probit model:
(1) Maximize a penalized log likelihood function, or
(2) Directly correct the estimator of
Advantages
For (1) estimates
i
so enables partial effects
Estimator is consistent under some circumstances
(Possibly) corrects in dynamic models
Disadvantage
No time invariant variables in the model
Practical implementation
Extension to other models? (Ordered probit model (maybe)
see JBES 2009)
A Mundlak Correction for the FE Model
*
it i
it it
i
y , i = 1,...,N; t = 1,...,T
y 1 if y > 0, 0 otherwise.
(Projection, not necessarily conditional mean)
w
i it it
i i
u
'
= o + + c
=
'
o = + +
Fixed Effects Model :
x
Mundlak (Wooldridge, Heckman, Chamberlain), ...
x
|
u
u 1 2
*
it
here u is normally distributed with mean zero and standard
deviation and is uncorrelated with or ( , ,..., )
y , i = 1,...,N; t = 1,..
i i i iT
i it it i
u
o
' '
= + + + c +
x x x x
Reduced form random effects model
x x u |
i
it it
.,T
y 1 if y > 0, 0 otherwise. =
Mundlak Correction
A Variable Addition Test for FE vs. RE
The Wald statistic of 45.27922 and
the likelihood ratio statistic of
40.280 are both far larger than the
critical chi squared with 5 degrees
of freedom, 11.07. This suggests
that for these data, the fixed
effects model is the preferred
framework.
Fixed Effects Models Summary
Incidental parameters problem if T < 10 (roughly)
Inconvenience of computation
Appealing specification
Alternative semiparametric estimators?
Theory not well developed for T > 2
Not informative for anything but slopes (e.g.,
predictions and marginal effects)
Ignoring the heterogeneity definitely produces an
inconsistent estimator (even with cluster correction!)
A Hobsons choice
Mundlak correction is a useful common approach.
Dynamic Models
'
= + + c +
'
= = + +
|
|
|
x
x
it it i,t 1 it i
it i,t 1 i0 it it i,t 1 i
y 1[ y u > 0]
Two similar 'effects'
Unobserved heterogeneity
State dependence = state 'persistence'
Pr(y 1| y ,..., y , x ,u] F[ y u ]
How to estimate , , marginal effects, F(.), etc?
(1) Deal with the latent common effect
(2) Handle the lagged effects:
This encounters the initial conditions problem.
Dynamic Probit Model: A Standard Approach
=
' = + +
[
T
i1 i2 iT i0 i i,t 1 i it
t 1
i1 i2 iT i0
(1) Conditioned on all effects, joint probability
P(y , y ,..., y | y , ,u ) F( y u , y )
(2) Unconditional density; integrate out the common effect
P(y , y ,..., y | y , )
i it
i
x x
x
=
' = o + u + o
}
i1 i2 iT i0 i i i0 i
2
i i0 i0 u i i1 i2 iT
i
P(y , y ,..., y | y , ,u )h(u | y , )du
(3) (The rabbit in the hat) Density for heterogeneity
h(u | y , ) N[ y , ], = [ , ,..., ], so
u =
i i
i i
x x
x x x x x x
' o + u +
=
' ' o + + + u + + o
[
}
i0 i it
i1 i2 iT i0
T
i,t 1 i0 u i it i i
t 1
y + w (contains every period of )
(4) Reduced form
P(y , y ,..., y | y , )
F( y y w , y )h(w )dw
This is a random effects model
i
i
it i
x x
x
x x
Simplified Dynamic Model
' = o + u + o
'
o + u +
i
2
i i0 i0 u
i i0 i
Projecting u on all observations expands the model enormously.
(3) Projection of heterogeneity only on group means
h(u | y , ) N[ y , ] so
u = y + w
(4) Re
i i
i
x x
x
=
' ' o + + + u + + o
[
}
i1 i2 iT i0
T
i,t 1 i0 u i it i i
t 1
duced form
P(y , y ,..., y | y , )
F( y y w , y )h(w )dw
Mundlak style correction with the initial value in the equation.
This is (again) a random effects mo
i
it i
x
x x
del
A Dynamic Model for Public Insurance
Age
Household Income
Kids in the household
Health Status
Basic Model
Add initial value, lagged value, group means
Dynamic Common Effects Model
1525 groups with 1 observation
were lost because of the
lagged dependent variable.
