Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
35 views

Homework #3 - Answers Economics 113 Introduction To Econometrics Professor Spearot Due Wednesday, October 29th, 2008 - Beginning of Class

This document contains the answers to homework questions for an economics class. It discusses interpreting slope coefficients in different regression specifications and whether examples satisfy assumptions of linear regression. Specifically: 1) It explains what the slope coefficient represents in terms of changes in the dependent and independent variables for different model specifications. 2) For examples of regressing exam grade on attendance and study hours, it notes the assumptions of zero conditional mean and random sampling are violated due to omitted variables and sample selection bias. 3) It states that omitting an important variable is the only option listed that can cause ordinary least squares estimates to be biased, as heteroskedasticity and correlation affect standard errors not parameter estimates.

Uploaded by

Cheung Tiffany
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
35 views

Homework #3 - Answers Economics 113 Introduction To Econometrics Professor Spearot Due Wednesday, October 29th, 2008 - Beginning of Class

This document contains the answers to homework questions for an economics class. It discusses interpreting slope coefficients in different regression specifications and whether examples satisfy assumptions of linear regression. Specifically: 1) It explains what the slope coefficient represents in terms of changes in the dependent and independent variables for different model specifications. 2) For examples of regressing exam grade on attendance and study hours, it notes the assumptions of zero conditional mean and random sampling are violated due to omitted variables and sample selection bias. 3) It states that omitting an important variable is the only option listed that can cause ordinary least squares estimates to be biased, as heteroskedasticity and correlation affect standard errors not parameter estimates.

Uploaded by

Cheung Tiffany
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Homework #3 Answers Economics 113 Introduction to Econometrics Professor Spearot Due Wednesday, October 29th, 2008 Beginning of class

ss 1. Please interpret the slope coefficient in each of the following four specifications: y = 0 + 1 x + y = 0 + 1 log( x) + log( y ) = 0 + 1 x + log( y ) = 0 + 1 log( x) + For the first specification:

y = 1 x
Thus, the slope coefficient equals the unit change in y after a unit change in x.

y = 1

x 1 x = *100 = 1 (percentage change in x ) x 100 x 100

Thus, the slope coefficient divided by 100 equals the unit change in y after a percentage change in x.
y = 1x y y * 100 = (1 * 100 )x y

Thus, the slope coefficient multiplied by 100 equals the percentage change in y resulting from a unit change in x.

y x = 1 y x y x * 100 = 1 *100 y x The slope coefficient gives the percentage change in y as a function of a percentage change in x.

2. For the following examples, discuss whether each satisfy the four assumptions we use for linear regression. If not, which assumptions are violated? a) To examine the link between attendance and grades, I construct an indicator variable, PickUp, which takes on the value of 1 if a student picked-up his or her exam on Friday, October 17th, and 0 otherwise. I then run the following regression using data for the entire class: ExamGrade = 0 + 1 PickUp + Here the problem is one of omitted variables. The indicator PickUp is likely correlated with a variable in the error term. For example, since studying probably improves your grade, and studying is likely correlated with your probability of attending class, there is bias in our estimates. In this case the assumption of zero conditional mean is violated.

b) To examine the link between studying and grades, I construct a new variable, Study, which is the self-reported hours-studied prior to the exam. The sampling was done on Friday, October 17th. Using this sample, I run the following regression: ExamGrade = 0 + 1 Study + You could make the same argument as in (a), instead using average attendance as the omitted variable. However, the bigger issue is that we are using a selected sample, so the random sampling assumption is violated. That is, we are only using the sample of those who attend class, which is likely not representative of the entire class. 3. Taken from Problem 3.7 in Wooldridge Which of the following can cause OLS estimators to be biased? (i) Heteroskedasticity (ii) Omitting an important variable (iii) A high correlation coefficient (say .95) between two random variables Only number two causes the estimators to be biased. Heteroskedasticity has to do with the estimates of error variance, which has nothing to do with the expected value of the parameter estimates. Further, high correlations only pose a problem for obtaining a precise estimate of variance.

You might also like