Scribd
Upload
49 views

Homework 2 Done

This document contains an analysis of CEO salary data using ordinary least squares regression. It begins by deriving the OLS estimator and showing it is an unbiased estimator of the population mean. It then analyzes CEO salary and tenure data, finding the average of each and information about the sample. Two models are estimated - salary as a function of tenure, and the log of salary as a function of tenure. Various regression outputs are interpreted and hypothesis tests are performed to analyze the relationship between salary and tenure. Stata code is included to replicate the analysis.

Uploaded by

mythien94
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
49 views

Homework 2 Done

This document contains an analysis of CEO salary data using ordinary least squares regression. It begins by deriving the OLS estimator and showing it is an unbiased estimator of the population mean. It then analyzes CEO salary and tenure data, finding the average of each and information about the sample. Two models are estimated - salary as a function of tenure, and the log of salary as a function of tenure. Various regression outputs are interpreted and hypothesis tests are performed to analyze the relationship between salary and tenure. Stata code is included to replicate the analysis.

Uploaded by

mythien94
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/ 4

ECON 7022: Homework 2

Thien Dong - GTID#: 903514870


September 24, 2019

1. Yi = α + ei , ei ∼ i.i.d.(0, σ 2 ), i = 1, 2, ..., T.
a) Derive the least squares estimator α̂.
T
(Yi − α̂)2 is minimized.
X
α̂ is chosen such that
i=1

Taking the first derivative with respect to α̂, we have:

T
X
(-2) (Yi − α̂) = 0
i=1
XT
⇐⇒ Yi − T.α̂ = 0
i=1
T
X
⇐⇒ α̂ = T1 . Yi = Ȳ .
i=1

b) Show that α̂ is unbiased.

α̂ is unbiased if E(α̂) = α.
From question a), we have that the least squares estimator α̂ is the sample mean Ȳ .
T T T
1
X
1
X
1
X 1
So E(α̂) = E(Ȳ ) = E( T . Yi ) = T E(Yi ) = T α = .T.α = α.
i=1 i=1 i=1 T

2. Use the dataset ceosal2.dta.

a) Find the average salary and the average tenure in the sample.
The average salary is 865,864.4 dollars, and the average tenure is 7.95 years.

b) There are 5 CEOs in their first year. The longest tenure in this sample is 37 years.

c) Estimate the model: salary = β0 + β1 .ceoten + textite


The β1 is interpreted as: with one additional year of tenure, the CEO would earn 11,746.13

1
dollars more in his or her salary.

The graph of yi versus xi :

The graph of ŷi versus xi (the fitted line):

2
The graph of eˆi versus xi :

d) Estimate the model: ln(salary) = β0 + β1 .ceoten + e


This β1 is interpreted as: an increase of one tenure year would result in (eβ1 − 1) × 100
percentage change in salary, which is approximately (e0.0097236 − 1) × 100 = 0.9771% change
in the salary.
The p-value for the hypothesis test H0 : β1 = 0 is 0.1284 (> 0.05). So we fail to reject H0 .
Using the t-test, we have:

βˆ1 − β1 0.0097236 − 0
t= = ≈ 1.53.
se(βˆ1 ) 0.0063645

At 95% confidence level, i.e., 5% significance level, the critical value tc = 1.9736 (with the
sample size T = 177 and 2 parameters, the degree of freedom is 177 - 2 = 175). Because t <
tc , again we fail to reject the null hypothesis that β1 = 0. The 95% confidence interval for β1
is [-0.0028374, 0.0222846], which includes 0.

3. Stata code
cd ”C:\mdong61\Dropbox (GaTech)\Computer\Documents\Semester I 2019 - 2020\Econometrics
I”
use ceosal2
summarize salary ceoten
count if ceoten == 0
reg salary ceoten
graph twoway scatter salary ceoten

3
graph twoway lfit salary ceoten
graph twoway (lfit salary ceoten) (scatter salary ceoten)
rvpplot ceoten
gen lnsalary = ln(salary)
reg lnsalary ceoten
test ceoten = 0
display invttail(175, 0.05/2)

You might also like