Scribd Logo
Upload

Welcome to Scribd!

  • 11 views

    Intecxass18 1

    Uploaded by

    Lukijan Mihajlovic
    This document outlines an introductory econometrics homework assignment. It includes 3 questions: 1) Analyzing simple linear regression and properties of estimators such as reverse regression and homogeneous regression. 2) Evaluating which issues in a multiple regression analysis on student GPA violating height and weight would violate Gauss-Markov conditions: student error variances correlated with name, multicollinearity between height and weight, and effect of weight-loss drug use on performance. 3) Showing that an alternative slope estimator using only the first two data points is unbiased but explaining why ordinary least squares is preferred over this alternative method.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Intecxass18 1

    Uploaded by

    Lukijan Mihajlovic
    11 views2 pages
    This document outlines an introductory econometrics homework assignment. It includes 3 questions: 1) Analyzing simple linear regression and properties of estimators such as reverse regression and homogeneous regression. 2) Evaluating which issues in a multiple regression analysis on student GPA violating height and weight would violate Gauss-Markov conditions: student error variances correlated with name, multicollinearity between height and weight, and effect of weight-loss drug use on performance. 3) Showing that an alternative slope estimator using only the first two data points is unbiased but explaining why ordinary least squares is preferred over this alternative method.

    Original Title

    intecxass18_1

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document outlines an introductory econometrics homework assignment. It includes 3 questions: 1) Analyzing simple linear regression and properties of estimators such as reverse regression and homogeneous regression. 2) Evaluating which issues in a multiple regression analysis on student GPA violating height and weight would violate Gauss-Markov conditions: student error variances correlated with name, multicollinearity between height and weight, and effect of weight-loss drug use on performance. 3) Showing that an alternative slope estimator using only the first two data points is unbiased but explaining why ordinary least squares is preferred over this alternative method.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    11 views2 pages

    Intecxass18 1

    Uploaded by

    Lukijan Mihajlovic
    This document outlines an introductory econometrics homework assignment. It includes 3 questions: 1) Analyzing simple linear regression and properties of estimators such as reverse regression and homogeneous regression. 2) Evaluating which issues in a multiple regression analysis on student GPA violating height and weight would violate Gauss-Markov conditions: student error variances correlated with name, multicollinearity between height and weight, and effect of weight-loss drug use on performance. 3) Showing that an alternative slope estimator using only the first two data points is unbiased but explaining why ordinary least squares is preferred over this alternative method.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    First homework assignment in

    Introductory Econometrics
    Robert M. Kunst
    October 15, 2016, answers to be submitted by October 22

    1. Simple linear regression

    (a) (Reverse regression) First consider a simple linear regression:

    yi = β0 + β1 xi + ui .

    We know the formula for its OLS slope estimate β̂1 and for the R2 .
    Older textbooks sometimes recommended running the ‘reverse re-
    gression’, with the roles of x and y exchanged, i.e.

    xi = γ0 + γ1 yi + vi .

    We wonder what we can learn from such a reverse regression. De-


    termine the formula for its OLS slope estimate γ̂1 . Show that the
    product of the two coefficients β̂1 γ̂1 equals the R2 and that this R2
    is identical in both regressions. [Hint: use yi − ȳ = β̂1 (xi − x̄) + ûi ]
    (b) (Homogeneous regression) Assume variables x, y follow the ‘inhomo-
    geneous’ standard regression scheme:

    yi = β0 + β1 xi + ui ,

    but you actually form an estimator assuming β0 = 0, i.e. you esti-


    mate the homogeneous regression

    yi = γxi + vi .

    Show that minimization of the sum of squared residuals for the ho-
    mogeneous regression
    ∑n
    (yi − γ̂xi )2
    i=1

    w.r.t. γ̂ really leads to the estimator


    ∑n
    yi xi
    γ̃ = ∑i=1
    n 2
    i=1 xi

    and try to evaluate the expectation E(γ̃|X) if the inhomogeneous


    model with β0 ̸= 0 holds. Note that the bias should depend on β0 .
    (c) In class, we have already shown that β̂1 can be written in the form
    ∑
    n
    xi − x̄
    β̂1 = β1 + wi ui , wi = ∑n ,
    i=1 (xi − x̄)
    2
    i=1

    such that the difference β̂1 − β1 is a weighted sum of the errors.


    We have also shown that the sum of the weights is 0. Further, we
    demonstrated that the two statistics β̂1 and ū are uncorrelated. A
    bit later, we considered the representation
    β̂0 = β0 + ū + (β̂1 − β1 )x̄.
    The two components ū, depending on the unweighted error sum, and
    β̂1 − β1 , depending on the weighted error sum, are uncorrelated, so
    the variance of the intercept estimate is the sum of two variances
    σ2 σ 2 (x̄)2
    var(β̂0 ) = + ∑n
    i=1 (xi − x̄)
    n 2

    This is still not the formula for the variance of the intercept estimate
    that appears on the slides
    ∑n
    σ2 x2
    var(β̂0 ) = ∑n i=1 i 2
    n i=1 (xi − x̄)
    Please show that the two formulas are identical.
    2. Something on multiple regression:

    Someone wants to regress the grade point average on the height and weight
    of n students. Let us assume that all Gauss-Markov conditions are fulfilled,
    except for one specific problem (such that for example problem (b) is
    not there for questions (a) and (c), which allows studying each problem
    independently). Please evaluate whether the following issues really violate
    the Gauss-Markov conditions and which one:
    (a) Students whose family name starts with a vowel have a 10% larger
    error variance than students whose name starts with a consonant;
    (b) Height and weight are strongly correlated, with a correlation coeffi-
    cient of around 0.9;
    (c) The students in the upper weight decile are known to have a high
    probability of abusing weight-loss drugs, the ingestion of which has
    a negative influence on student performance.
    3. Again consider simple regression. We know that under the first four Gauss-
    Markov assumptions OLS is unbiased. Now consider the following ‘alter-
    native’ slope estimator, assuming the sample size n > 2:
    y2 − y1
    β̃1 =
    x2 − x1
    Show that this estimator is unbiased under the Gauss-Markov assump-
    tions. Why would we rather not use this alternative estimator and prefer
    OLS?

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy