ECMT5001: Final Examination (2022s1)
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ECMT5001: Final Examination (2022s1)
1. [Total: 20 marks] Bob, the proud owner of ìHungry Bob,î wanted to study how the fast food restaurantís revenue is related to customersí satisfaction. He collected data on daily revenue R (in $1,000) and customer satisfaction score S (in percentage points) over time, and ran a simple linear regression of R on S . He obtained the following regression result (standard error in parenthesis).
R^ = 2:1 + 0:5S:
(1:0) (0:3)
(a) [3 marks] Interpret the intercept term.
(b) [3 marks] Interpret the slope coe¢cient.
(c) [4 marks] Let D denote the customer satisfaction score recorded in decimal points, i.e., D = S . What is the estimation result of a regression of R on D based on the same data? Your answer should be in the following form (standard error in parenthesis):
R^ = ? + ? D:
(?) (?)
(d) A competing restaurant ìBurger Queenî was opened next door. Let I denote the time dummy which equals 1 when Burger Queen was open, and 0 otherwise. To take this into account, Bob considered the regression as follows:
R = &0 + &1S + &2I + &3S ! I + u:
Using a sample spanning 80 days, Bob estimated the regression. The result is dis- played below (standard error in parenthesis):
R^ = 2:5 + 0:4S " 0:7I +0:25S ! I:
(0:8) (0:3) (0:3) (0:1)
i. [5 marks] Bob suspected that the presence of Burger Queen is associated with a drop in his revenue. Test Bobís suspicion at the 1% signiÖcance level. Show all your steps.
ii. [5 marks] Bob claimed that the presence of Burger Queen a§ects the relationship between Hungry Bobís revenue and customer satisfaction score. Test Bobís claim at the 1% signiÖcance level. Show all your steps.
2. [Total: 20 marks] Baobao ran the following simple linear regression of y on x y = &0 + &1x + u:
He obtained the following summary statistics from the sampled data:
= 0:5, y+ = 0:4, n = 100
100
X(xi " )2 = 0:2
i=1
100
X(yi " y+)2 = 0:3
i=1
100
X(xi " )(yi " y+) = 0:1
i=1
(a) Compute:
i. [3 marks] 1 (OLS estimator of &1 )
ii. [3 marks] 0 (OLS estimator of &0 )
iii. [3 marks] regression R2
iv. [3 marks] SSR (sum of squared residuals)
v. [3 marks] standard error of 1
(b) [4 marks] Suppose the error series is negatively serially correlated. Will the serial correlation of the error series result in a biased OLS estimator 1 for & 1 ? Explain.
3. [Total: 20 marks] Carol is studying the relationship between the number of cars (ncar) and the air quality index (AQI) across a number of countries. She considers the following regression:
AQI = &0 + &1ncar + &2 (ncar)2 + u: (1)
She also ran the following auxiliary regression
2 = 00 + 01ncar + 02(ncar)2 + v; (2)
where is the residual from regression (1).
The following information was extracted from the ANOVA table associated with the aux- iliary regression :
Regression R2 |
0.28 |
Number of observations |
55 |
Sum of squared residuals |
12.31 |
(a) [6 marks] With the help of the auxiliary regression (2), conduct a test for het-
eroskedastic errors in regression at the 5% signiÖcance level. Show all your steps.
(b) [4 marks] Suppose the errors in regression (1) are heteroskedastic. Comment on the
accuracy of the OLS estimates of &0 , & 1 and &2 and their OLS standard errors.
(c) Carol thought that it is reasonable to assume that Var(ujncar) = 42 $ (ncar)2 for the errors in regression (1).
i. [6 marks] Explain how Carol can transform regression (1) so that the regression errors become homoskedastic.
ii. [4 marks] Describe an alternative method which Carol may use to conduct valid statistical inference on the parameters in regression (1).
2022-11-10