Scribd
Upload
6 views9 pages

DRAFT_13 Multiple Linear Regression_Further Issues

The document outlines key concepts in multiple linear regression, focusing on functional forms, adjusted R-squared, and nested versus non-nested models. It discusses various functional forms including log forms, quadratic functions, higher-order polynomials, and interaction terms. Additionally, it provides formulas for adjusted R-squared and examples for model comparison.

Uploaded by

angagalac1
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
6 views9 pages

DRAFT_13 Multiple Linear Regression_Further Issues

The document outlines key concepts in multiple linear regression, focusing on functional forms, adjusted R-squared, and nested versus non-nested models. It discusses various functional forms including log forms, quadratic functions, higher-order polynomials, and interaction terms. Additionally, it provides formulas for adjusted R-squared and examples for model comparison.

Uploaded by

angagalac1
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/ 9

Econ 131 – Quantitative Economics

(a.k.a. Introductory Econometrics)


Multiple linear regression: Further issues
Instructor: Anthony G. Sabarillo
University of the Philippines School of Economics
Main reference: Wooldridge (2012), Prof. Jandoc’s lecture notes

Not for distribution. For lecture purposes only. 1


Outline
• More functional forms
• Adjusted 𝑅𝑅2

Not for distribution. For lecture purposes only. 2


Functional forms
• Some remarks on log forms:
• Suppose the population model is:
• log 𝑦𝑦 = 𝛽𝛽0 + 𝛽𝛽1 log 𝑥𝑥1 + 𝛽𝛽2 𝑥𝑥2 + 𝑢𝑢
• log 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = 𝛽𝛽0 + 𝛽𝛽1 log 𝑛𝑛𝑛𝑛𝑛𝑛 + 𝛽𝛽2 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 + 𝑢𝑢
• ⇒ log � 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = 9.23 + 0.718 log 𝑛𝑛𝑛𝑛𝑛𝑛 + 0.306 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 + 𝑢𝑢
• Δ𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 1 ⇒ 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ↑ by approximately 30.6%.
• Recall: % Δ𝑦𝑦 ≈ 100 × Δ log 𝑦𝑦 %
• As Δ log 𝑦𝑦 becomes larger and larger, the approximation becomes less accurate
• To get the exact % Δ�y using 𝛽𝛽̂2 , use this formula:
Δ𝑦𝑦� �
• = 𝑒𝑒 𝛽𝛽2Δ𝑥𝑥2 − 1
𝑦𝑦 0
Δ𝑦𝑦�
• What is when Δ𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 1 and when Δ𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = −1?
𝑦𝑦 0

Not for distribution. For lecture purposes only. 3


Example:
Functional forms � = 3.73 + .298𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 − .0061𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑟𝑟 2
𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤
(.35) (.041) (.0009)
• Another way of handling nonlinearities: 2
𝑛𝑛 = 526, 𝑅𝑅 = 0.093
Quadratic functions

• 𝛽𝛽0 > 0, 𝛽𝛽1 < 0, 𝛽𝛽2 > 0 • 𝛽𝛽0 , 𝛽𝛽1 > 0, 𝛽𝛽2 < 0
• At first, the marginal effect decreases • At first, the marginal effect increases
at a decreasing rate. at a decreasing rate.
• Beyond the threshold, it increases • Beyond the threshold, it decreases
at an increasing rate. at an increasing rate.
Not for distribution. For lecture purposes only. 4
Functional forms
• Another way of handling nonlinearities:
Higher-order polynomials

• 𝛽𝛽0 , 𝛽𝛽1 , 𝛽𝛽3 > 0, 𝛽𝛽2 < 0 • 𝛽𝛽0 , 𝛽𝛽1 , 𝛽𝛽2 > 0, 𝛽𝛽3 < 0

Not for distribution. For lecture purposes only. 5


Functional forms
• Another way of handling nonlinearities:
Reciprocal functions

• 𝛽𝛽0 , 𝛽𝛽1 > 0 • 𝛽𝛽0 > 0, 𝛽𝛽1 < 0

Not for distribution. For lecture purposes only. 6


Functional forms
• What if the effect of 𝑥𝑥𝑗𝑗 depends on the level of yet another explanatory variable?
• Functional forms with interaction terms
• 𝑦𝑦 = 𝛽𝛽0 + 𝛽𝛽1 𝑥𝑥1 + 𝛽𝛽2 𝑥𝑥2 + 𝛽𝛽3 𝑥𝑥1 ∗ 𝑥𝑥2 + 𝑢𝑢
𝜕𝜕𝜕𝜕(𝑦𝑦𝑦𝐱𝐱)
• ⇒ = 𝛽𝛽1 + 𝛽𝛽3 𝑥𝑥2
𝜕𝜕𝑥𝑥1

𝜕𝜕𝜕𝜕(𝑦𝑦𝑦𝐱𝐱)
• ⇒ = 𝛽𝛽2 + 𝛽𝛽3 𝑥𝑥1
𝜕𝜕𝑥𝑥2

• Usually, 𝛽𝛽2 (the partial effect of 𝑥𝑥2 on 𝑦𝑦 when 𝑥𝑥1 = 0) is not of interest.
• How do we reparametrize the model above so that we can estimate a coefficient that
gives the partial effect of 𝑥𝑥2 on 𝑦𝑦 at the mean value of 𝑥𝑥1 ?

Not for distribution. For lecture purposes only. 7


Adjusted 𝑅𝑅2
• Formula:
� 2 = 1 − 𝑆𝑆𝑆𝑆𝑆𝑆/(𝑛𝑛−𝑘𝑘−1)
• 𝑅𝑅
𝑆𝑆𝑆𝑆𝑆𝑆/(𝑛𝑛−1)

• Alternative:
1−𝑅𝑅2 𝑛𝑛−1
• 𝑅𝑅� 2 =1−
𝑛𝑛−𝑘𝑘−1

• Note: 𝑅𝑅� 2 < 0 is possible, which indicates a poor model fit relative to the
number of degrees of freedom.

Not for distribution. For lecture purposes only. 8


Nested and non-nested models
• Which of the two pairs of equations are nested models?
• 𝑦𝑦 = 𝛽𝛽0 + 𝛽𝛽1 𝑥𝑥1 + 𝛽𝛽2 𝑥𝑥2 + 𝛽𝛽3 𝑥𝑥3 + 𝑢𝑢
• 𝑦𝑦 = 𝛽𝛽0 + 𝛽𝛽1 𝑥𝑥1 + 𝛽𝛽2 𝑥𝑥2 + 𝑣𝑣

• 𝑦𝑦 = 𝛽𝛽0 + 𝛽𝛽1 𝑥𝑥1 + 𝛽𝛽2 𝑥𝑥2 + 𝛽𝛽3 𝑧𝑧1 + 𝑢𝑢


• 𝑦𝑦 = 𝛼𝛼0 + 𝛼𝛼1 𝑥𝑥1 + 𝛼𝛼2 𝑥𝑥2 + 𝛼𝛼3 𝑧𝑧2 + 𝑣𝑣

• Which model should you pick? (Justify your answer):


• 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝛽𝛽𝛽 + 𝛽𝛽1 log 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑢𝑢, 𝑅𝑅2 = 0.06, 𝑅𝑅� 2 = 0.03
• 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 = 𝛼𝛼0 + 𝛼𝛼1 𝑠𝑠𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 + 𝛼𝛼2 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 2 + 𝑢𝑢, 𝑅𝑅2 = 0.148, 𝑅𝑅� 2 = 0.09
Not for distribution. For lecture purposes only. 9

You might also like