Scribd
Upload
42 views2 pages

Exercises2 2022 2023v1

1. The document is an exercise sheet on discrete choice models from a PhD course in econometrics. 2. It presents two exercises on binary choice models - the first models a latent variable that is observed as 1 if below a threshold and 0 if above, and the second models the probability of an outcome as a function of covariates using a logistic function. 3. The exercises ask students to derive maximum likelihood estimators, show properties like invariance to transformations, and apply a likelihood ratio test to hypothesis testing.

Uploaded by

Ermita Yusida
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
42 views2 pages

Exercises2 2022 2023v1

1. The document is an exercise sheet on discrete choice models from a PhD course in econometrics. 2. It presents two exercises on binary choice models - the first models a latent variable that is observed as 1 if below a threshold and 0 if above, and the second models the probability of an outcome as a function of covariates using a logistic function. 3. The exercises ask students to derive maximum likelihood estimators, show properties like invariance to transformations, and apply a likelihood ratio test to hypothesis testing.

Uploaded by

Ermita Yusida
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

University of Lisbon ISEG

2022/2023 Paulo M.D.C. Parente

Advanced Econometrics
PhD in Economics
2022/2023
Exercise Sheet 2 - Discrete Choice Models
(version 1/10/2022)

1. Consider a latent variable modeled by Yi = Xi0 + "i , with "i N (0; 1). Suppose we
observe only Yi = 1 if Yi < Ui and Yi = 0 if Yi Ui , where the upper limit Ui is a
known constant for each individual (i.e., data) and may di¤er over individuals.

(a) Find P[Yi = 1jXi ; Ui ]. [Hint: Note that this di¤ers from the standard case both
due to presence of Ui and because the equalities are reversed with Yi = 1 if
Yi < Ui .]
(b) Provide details on an estimation method to consistently estimate .

2. Consider a binary random variable Yi such that


P (Yi = 1jDi ) = G ( + Di ) ; i = 1; :::; n;
where Di is a dummy variable and G ( ) is a strictly increasing known function that
satis…es 0 < G ( ) < 1: Assume that the sample f(Yi ; Di )gni=1 corresponds to a random
sample and denote nij the number of elements of the sample that satisfy Y = i and
D = j, where i = 0; 1 and j = 0; 1: Assume initially that nij is strictly positive for all
i and j.

(a) Show that the maximum likelihood estimators of and are given by

1 n10
^ = G ;
n10 + n00
^ = G 1 n11 1 n10
G ;
n01 + n11 n10 + n00
where G 1 ( ) is the inverse function of G ( ) [Hint: Maximize the log-likelihood
function with respect to A = G( ) and B = G( + ) and use the invariance of
maximum likelihood estimators to transformations to obtain ^ and ^ :]
(b) Show that the log-likelihood function evaluated at ^ and ^ does not depend on
G( ):
(c) Prove that the maximum likelihood estimator does not exist if n10 = 0.
(d) The data consist of 100 observations as follows:
Y
0 1
D 0 24 28
1 32 16
and assume that G (z) = exp (z) = [1 + exp (z)] :

1
University of Lisbon ISEG
2022/2023 Paulo M.D.C. Parente

i. Obtain the maximum likelihood estimates of and .


ii. Test the hypothesis that = 0 by using a likelihood ratio test.

You might also like