Scribd
Upload
162 views3 pages

Thanh Huyen - Week 5 - Final Test Answer

This document discusses using a logit model to analyze responses to a direct mailing. It provides the maximum likelihood estimates of the model parameters. It then defines the marginal effect and elasticity of active status. It shows how to compute these values for a 50-year-old active and inactive male customer. It further defines the elasticity in terms of the probability of response given active/inactive status. It proves this alternative definition can be simplified and uses it to compute the elasticity for a 50-year-old active male customer.

Uploaded by

Nguyễn Sapphire Thanh Huyền
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
162 views3 pages

Thanh Huyen - Week 5 - Final Test Answer

This document discusses using a logit model to analyze responses to a direct mailing. It provides the maximum likelihood estimates of the model parameters. It then defines the marginal effect and elasticity of active status. It shows how to compute these values for a 50-year-old active and inactive male customer. It further defines the elasticity in terms of the probability of response given active/inactive status. It proves this alternative definition can be simplified and uses it to compute the elasticity for a 50-year-old active male customer.

Uploaded by

Nguyễn Sapphire Thanh Huyền
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Consider again the application in lecture 5.

5, where we have analyzed response to a direct


mailing using the following logit specification
exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )
Pr[respi = 1] = 1 + exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )

for i = 1, : : : , 925. The maximum likelihood estimates of the parameters are given by

(a) The marginal effect of activity status is defined as


∂Pr[resp𝑖 = 1]
= Pr[respi = 1] Pr[respi = 0]β2.
∂active𝑖

We could use this result to construct an activity status elasticity


∂Pr[resp𝑖 = 1] active𝑖
= Pr[respi = 0]activeiβ2.
∂active𝑖 Pr[resp𝑖 = 1]
Use this results to compute the elasticity effect of active status for a 50 years old active male
customer.
Do the same for an 50 years old inactive male customer.

Answer:

For a 50 years old active male customer:


+ male = 1
+ active = 1
+ age = 50

exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )


So we have: Pr[respi = 1] =
1 + exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )

exp(−2.488 + 0.954x1 + 0.914x1 + 0.07x50 + (−0.069)x(50/10)2 )


=
1 + exp(−2.488 + 0.954x1 + 0.914x1 + 0.07x50 + (−0.069)x(50/10)2 )

3.174023
= = 0.760423
4.174023

 Pr[respi = 0] = 1 - Pr[respi = 1] = 1- 0.760423 = 0.239577


 Elasticity = Pr[respi = 0] x activei x β2 = 0.239577 x 1 x 0.914 = 0.218973
For a 50 years old inactive male customer:
Elasticity = Pr[respi = 0] x activei x β2 = 0.239577 x 0 x 0.914 = 0

(b) The activity status variable is only a dummy variable and hence it can take only two values. It
is therefore better to define the elasticity as

Pr[resp𝑖 = 1|active𝑖 = 1] − Pr[resp𝑖 = 1|active𝑖 = 0]


Pr[resp𝑖 = 1|active𝑖 = 0]
Show that you can simplify the expression for the elasticity as
(exp(β2) - 1) Pr[respi = 0|activei = 1].

Answer:

Let’s assume that:

age𝑖 2
exp (β0 + β1male𝑖 + β3age𝑖 + β4 ( ) )=A
10

exp(β2active𝑖 ) = B

Pr[resp𝑖 = 1|active𝑖 = 1]− Pr[resp𝑖 = 1|active𝑖 = 0]


=
Pr[resp𝑖 = 1|active𝑖 = 0]
2 2
exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10) ) exp(β0 + β1male𝑖 + β3age𝑖 + β4(age𝑖/10) )
2
− 2
1 + exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10) ) 1 + exp(β0 + β1male𝑖 + β3age𝑖 + β4(age𝑖/10) )
2
exp(β0 + β1male𝑖 + β3age𝑖 + β4(age𝑖/10) )
2
1 + exp(β0 + β1male𝑖 + β3age𝑖 + β4(age𝑖/10) )
AB AB A 1 + A
− A ( − )𝑋 𝐴𝐵(1+𝐴) 𝐵−1
= 1 + ABA 1 + A = 1 + ABA 1 +1A+ A A = 𝐴(1+𝐴𝐵) − 1 = 1+𝐴𝐵
𝑋
1+A 1+A A

exp(β2active𝑖 )−1
= (*)
1+exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )

1
Since = Pr[respi = 0|activei = 1].
1+exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )

(*) = (exp(β2) - 1) Pr[respi = 0|activei = 1]

(c) Use the formula in (b) to compute the activity elasticity of 50 years old male active
customer.

For a 50 years old active male customer:


+ male = 1
+ active = 1
+ age = 50

Elasticity = (exp(β2) - 1) Pr[respi = 0|activei = 1]

exp(β2active𝑖 )−1
=
1+exp(β0 + β1male𝑖 + β2active𝑖 + β3age𝑖 + β4(age𝑖/10)2 )

exp(0.914x 1 )−1
=
1+exp(−2.488 + 0.954x1 + 0.914x1 + 0.07x50 + (−0.069)x(50/10)2 )

= 0.357995

You might also like