You are on page 1of 25

Empirical Methods for

Microeconomic Applications

William Greene
Department of Economics
Stern School of Business
Lab 4. Ordered Choice and Count Data
Upload Your Project File
Binary Dependent Variables
DOCTOR = visited the doctor at least once
HOSPITAL = went to the hospital at least once.
PUBLIC = has public health insurance (1=YES)
ADDON = additional health insurance.(1=Yes)

ADDON is extremely unbalanced.


Ordered Dependent Variable

HSAT = ordered reported health


satisfaction, coded 0,1,,10.

Use with ORDERED


or ORDERED ; Logit
or
OPROBIT and OLOGIT
HISTOGRAM ; Rhs = hsat $
HISTOGRAM ; Rhs = hsat ; Group = female
; Labels = Male,Female
; Title = Health Satisfaction by Gender $
Ordered Choice Models
Ordered ; Lhs = dependent variable
; Rhs = One, independent variables $

Remember to include the constant term


For ordered logit i stead of ordered probit, use

Ordered ; Logit ; Lhs = dependent variable


; Rhs = One, independent variables $

To get marginal effects, use ; Margin as usual.

There are fixed and random effects estimators for this model:
; FEM ; Panel
; Random ; Panel
Sample Selection in Ordered Choice
Selection Model - The usual binary choice model
z* = z+w
z = 1 if z* > 0
When (and only when) z = 1, we observe the ordered probit.
y* =
x + e, we assume x contains a constant term
y = j if m j-1 < y* m j , j = 0,1,...,J
m-1 = -, m o = 0, m J = +, m j-1 < m j, j = 1,...,J
If r=Cor(w,e) 0, the ML estimator of the ordered probit
model is inconsistent.
Sample Selection Ordered Probit

PROBIT ; Lhs = ; Rhs = ; HOLD $


ORDERED ; Lhs = ; Rhs = ; Selection $

This is a maximum likelihood estimator, not a least


squares estimator. There is no lambda variable.
The various parameters are present in the
likelihood function.
PROBIT;if[year1994=1]; lhs=public;rhs=one,age,income,working ;hold$
ORDERED; If[year1994=1] ; lhs=health;rhs=x;selection$
Hierarchical Ordered Probit
Hierarchical ordered probit. Ordered probit in which threshold
parameters depend on variables. Two forms:

HO1: (i,j) = exp[(j) + z(i)].


HO2, different vector for each j.

Use ORDERED ; ; HO1 = list of variables or


ORDERED ; ; HO2 = list of variables.

Can combine with SELECTION models and ZERO INFLATION models.

This is also the Pudney and Shields generalized ordered probit from
Journal of Applied Econometrics, August 2000, with the
modification of using exp() and internally, a way to make sure
that the thresholds are ordered..
Count Data
DOCVIS = count of visits to the doctor
HOSPVIS = count of visits to the hospital.
There are outliers. It helps to use truncated or censored samples
(1) Truncated Data
SAMPLE ; All $
MODEL ; If [ docvis <= 10] ; Lhs = $
(2) Censored Data
SAMPLE ; All $
CREATE ; DocVis10 = Min(10,DocVis) ; Hosp10 = Min(10,HospVis) $
There are also excess zeros. ZIP or HURDLE model suggested.
Models for Count Data
Basic models Poisson and negative binomial
POISSON ; Lhs = y ; Rhs = One, $
NEGBIN ; Lhs = y ; Rhs = One, $

NEGBIN ; Model=NB1 or NB2 or NBP ;

Many extensions
Various heterogeneity forms
Panel data
Random parameters and latent class
Zero inflation
Sample selection
Censoring and truncation
Numerous others (some, far from all, shown below)
NEGBIN ; If[year1994=1];Lhs=docvis;rhs=x;model=nbp$
Sample Selection
Selection Model - The usual binary choice model
z* = z+w
z = 1 if z* > 0
When (and only when) z = 1, we observe the Poisson or
Negative binomial outcome variable and covariates.
y = 0,1,... with probability
exp[-l(e)][l(e)] j
Prob(y=j|x,z=1,e)=
j!
l(e) = exp(
x + e)
" Selectivity " arises if Cor(w,e)= r 0.
The model is fit by full information maximum likelihood, using
Hermite quadrature to integrate e out of the likelihood.
Selection Model
Probit ; Lhs = ; Rhs = ; Hold $
Poisson ; Lhs = ; Rhs = ; Selection ; MLE$
Zero Inflation
Two regime, latent class model
Prob[Regime 1 => y=0] = q
Prob[y = j|Regime 2] = Poisson or NegBin,
=exp(x)
Reduced form:
Prob[y=0] = q + (1-q)P(0)
Prob[y=j > 0] = (1-q)P(j)
Regime Models:
q = Probit or Logit
Structures:
ZIP: Probit or Logit F(z) z can be any set of variables
ZIP-tau: Probit or Logit F( x) same x as above
ZIP and ZIP-tau Models
;ZIP Logit, ZIP-tau
;ZIP = Normal: Probit ZIP-tau
;ZIP [=Normal] ; Rh2 = variables in z
Alternative Models
Default is Poisson
; MODEL = NegBin
; MODEL = Gamma
Hurdle Model
Two Part Model:
Prob[y=0] = Logit or Probit using x from the
count model or z as specified with ;RH2=list
Prob[y=j|j>0] = Truncated Poisson or NegBin

Two part decision:


POISSON ; ; Hurdle $
POISSON ; ; Hurdle ; Rh2 = List $
POISSON ; If[year1994=1]; Lhs = docvis;rhs=x ; hurdle
; Rh2 = one,age,income,working $
PARTIALS ; If[year1994=1] ; effects: x ; summary $
Poisson and NB Mixture Models
with Normal Heterogeneity
y = 0,1,... with probability
exp[-l(e)][l(e)]j
Prob(y=j|x,z=1,e)=
j!
l(e) = exp(
x + e)
The model is fit by full information maximum likelihood, using
Hermite quadrature to integrate e out of the likelihood. Note,
this is precisely the negative binomial model if exp(e) has a
aa a-1
gamma distribution with mean 1; f[exp( e)]= e exp( -ae).
G(a)
In this model, e has a normal distribution, so exp(e) is lognormal.
Add ; NORMAL to the command.

You might also like