Scribd
Upload
77 views

(Open The Confidence Interval Calculator Excel File On Moodle. The

This document provides exercises on introductory econometrics. It includes: 1. Estimating a regression model using OLS and calculating R-squared and other statistics. 2. Using the regression results to calculate a point prediction, describe the distribution of the estimator, and find a confidence interval for the prediction. 3. Describing the distribution of the prediction error and finding a confidence interval. 4. Conducting hypothesis tests to test if the population regression line passes through the origin, if the slope is zero, and if the slope is negative as an a priori belief.

Uploaded by

patientbioinvest
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
77 views

(Open The Confidence Interval Calculator Excel File On Moodle. The

This document provides exercises on introductory econometrics. It includes: 1. Estimating a regression model using OLS and calculating R-squared and other statistics. 2. Using the regression results to calculate a point prediction, describe the distribution of the estimator, and find a confidence interval for the prediction. 3. Describing the distribution of the prediction error and finding a confidence interval. 4. Conducting hypothesis tests to test if the population regression line passes through the origin, if the slope is zero, and if the slope is negative as an a priori belief.

Uploaded by

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

EC2017: Introductory Econometrics

Week 4 Exercises
[Open the Confidence Interval Calculator Excel file on moodle. The
default data is exactly the one used in the exercise below. I think you
should use these two teaching aids in conjunction.)
Consider the following dt on !ge "left column# nd Wge "right column# "used in the first week of
lectures#
!ge "x# Wge "y#
21 150
25 230
31 356
17 103
24 500
$ou now h%e estimted the model yi & + xi+ i 'y fitting the line y
e
& ('x & )1*1+,-
( 1*+4-x using ./0
i+ 1ind 2
2
nd 2esidul 0um of 03ures "200#
TSS = (y- )
2
= 104160.8
ESS = b
2
(x- )
2
= 19.!
2
"107.2 =#!1.$
So %SS = 63546.07
R
2
=#!1.$&104160.8=0.3899
ii+ In%oke the ssum4tions of the C/25 to find the following:
"# # 1or x & 2,6 clculte 4oint estimte of the 4redictor for the corres4onding y+
y
e
= -191.56+19.46*25 = 294.94 is the point estimate of the predictor
"'# Wht is the distri'ution of the estimtor used in 4rt "# 'o%e7
Since y
e
= a + bx !"y
e
# = !"a# + x*!"b# for any $i%en %a&'e of x. So
!"y
e
# = !"a# + x*!"b# = + x = !"y# beca'se both a and b are 'nbiased estimators.
(&so since both a and b ha%e norma& distrib'tions and y
e
is a &inear f'nction of a and b
y
e
has a norma& distrib'tion as )e&& )ith %ariance % $i%en by
%
2
* "1+n# + "x2, - x2-!(.#
2
+(x2i - x2-!(. #
2
/
21102.,2*"1+5# + "25-21.6#
2
+1,2.2/ = 4621.69
3
[.ote4 5ere remember that as disc'ssed in the &ect'res x is the same as x2 and
x2-!(. = ' so that x2, = 25 is the ne)&y ac6'ired %a&'e of x for )hich the act'a& y
is 21, as per the data $i%en./
7h's "y
e
8 !"y##+ "4621.69#
1+2
has a t-distrib'tion )ith 1 de$rees of freedom
"c# 8se the distri'ution in 4rt "'# to clculte the *,9 confidence inter%l for this
4rediction+
Since "y
e
8!" y##+ "4621.69#
1+2
has a t-distrib'tion )ith 1 de$rees of freedom and the
959 %a&'es for t1 are 1.1024 the 959 confidence inter%a& for the predictor is =
*294.94 - 60*1.1024 294.94+60*1.1024/ = *20.54 511.14/
"d# Wht is the error of 4rediction nd its distri'ution for the gi%en %lue of x & 2,7
Wht is the *,9 confidence inter%l for the error of 4rediction7
7he prediction error p = y- y
e
is norma&&y distrib'ted beca'se both y
e
and y are. (&so p
c&ear&y has :ero mean "pro%ed in the &ect'res#. 7he %ariance of p is $i%en by
%
2
* 1+ "1+n# + "x2, - x2-!(.#
2
+(x2i - x2-!(. #
2
/ 250,5.2. 7herefore p+"250,5.2
#
1+2
has a t distrib'tion )ith 1 de$rees of freedom.
Since p+"250,5.2#
1+2
has a t distrib'tion )ith 1 de$rees of freedom and the re&e%ant t
%a&'es are 1.1024 )e can )or; o't the 959 confidence inter%a& of p as 1.1024*
"250,5.2#
1+2
or 511.226

:y4othesis ;esting:
"# ;est "t ,9 significnce# the hy4othesis tht the 4o4ultion regression line 4sses
through the origin+
5,4 = ,< 514 ,. =onstr'ct t = -191.56+ 110.,652 ,.52 > critica& t = 1.1024"at
59 t)o-tai&# . So do not re?ect 5, and retain the hypothesis that the &ine passes thro'$h
ori$in.
"'#;est "t ,9 significnce# the hy4othesis tht the slo4e coefficient is <ero+
5,4 = ,< 514 ,. =onstr'ct t = 19.46+14.,6 1.10 > critica& t = 1.1024. So do not
re?ect 5, and retain the hypothesis that the s&ope coefficient is :ero the mode& has no
exp&anatory po)er.
"'# ( priori resercher 'elie%es tht the slo4e coefficient is negti%e num'er+ :ow do
they construct hy4othesis test to %erify this7 [:int: drw digrm for the t)
distri'ution#
5,4 = ,< 514 > ,. =onstr'ct t = 19.46+14.,6 1.10 t @ -2.15 "at 59 one-tai&#. So do
not re?ect 5, and retain the hypothesis that the s&ope coefficient is :ero not ne$ati%e.
.-.!" .#.!$
t=
%

You might also like