ECON20003 QUANTITATIVE METHODS 2 2020
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
DEPARTMENT OF ECONOMICS
SUMMER SEMESTER ASSESSMENT, 2020
ECON20003 QUANTITATIVE METHODS 2
PART A
A1. True/False
Note:
• Answer each question below (1, 2, 3, 4) by writing (in full) either “True” or “False” in the answer booklet; no marks will be given to answers written on this question paper.
• No explanation of your choice (True or False) is required. However, if you have made the wrong choice, partial credit may be given for a worthy explanation or useful working.
For testing whether a population mean (1) is at least as high as another population
mean (2), the appropriate alternative hypothesis is 2 > 1 .
When testing whether a population mean () is greater than 5 based on a sample mean () of 4, the -value for the test will be greater than 50%.
Estimation of a simple regression model for a binary dependent variable yields intercept and slope estimates of 1.1871 and –0.1022, respectively. If the predicted probability of the dependent variable taking a value of one when the independent variable takes a value of eight is 35.19% (i.e., r( = 1| = 8) = 0.3519), the estimation method must have been ordinary least squares (OLS).
If − −1 = 0 + 1 −1 + , where is an error term, it follows that is stationary if −2 < 1 < 0.
A2. Short Calculation Questions
1. A winemaker asks 100 customers to rate each of his shiraz and merlot red wines on the following scale: 1 = Bad; 2 = OK; 3 = Good. Ofthe 100 customers, 40 rated the shiraz higher, and the rest were equally split between rating the wines the same and rating the merlot higher. Based on this sample information, having rounded the calculated test statistic to two decimal places, find the -value for testing the null hypothesis that the wines are equally preferred against the alternative that the shiraz is preferred.
2. Independent random samples of 120 men and 100 women were asked if they illegally
download digital product (e.g., a movie or music). One in five women admitted to
this sample information, having rounded the calculated test statistic to two decimal places, find the -value for testing the null hypothesis that an equal proportion ofmen and women illegally download against the alternative that a higher proportion of women illegally download.
3. Given the (partially hidden) EViews output below, calculate a 90% confidence interval for the slope coefficient.
Dependent Variable: Y
Included observations: 8000
Variable |
Coeff. |
Std. Error |
Z-Statistic |
Prob. |
C X |
0.1132 0.0766 |
0.3017 |
0.3752 |
0.7075 0.1010 |
4. Suppose that running a multiple regression yields
= −17.73 + 2.58 + 0.20
and that running a simple regression of on using the same data set yields = −10.40 + 2.40 .
Calculate the slope parameter estimate that would be obtained from running a simple
regression of on using the same data set; i.e.,
= 0 + 1 + .
PART B
B1. Ordinal data
A lecturer teaches the same intermediate statistics course at three different colleges in Melbourne; colleges A, B and C. The lecturer wants to test whether student perceptions about the usefulness of the course differ between colleges. The lecturer surveyed six randomly selected students from each college who completed the course last year and asked them to rate the usefulness of the course based on the following scale:
1 = not very useful; 2 = quite useful; 3 = very useful.
In college A, all six students gave a rating of 1. In college B, three students gave a rating of 1
and three gave a rating of 2. In college C, three gave a rating of2 and three gave a rating of 3.
(a) Given the sample results described above, calculate the rank sum for college A.
(b) Given the sample results described above, calculate the rank sums for colleges B and C.
(c) Test whether student perceptions of the course’s usefulness differ between colleges at the 5% significance level and state your conclusion.
(d) To investigate gender differences, the lecturer asked 20 randomly selected male students and 20 randomly selected female students from one of the colleges to rate the course’s usefulness based on the three-point scale described above (1 = not very useful, …, 3 = very useful). If the lecturer appropriately concluded at the 5% significance level that males found the course more useful than females based on the large-sample approximation of the Wilcoxon rank sum test, below what whole numerical value must the female students’ rank sum have been?
B2. Regression
CardigANZ is a company that manufactures high-quality woollen cardigans that it sells in its own retail stores across Australia and New Zealand. In any given store, all cardigans sell for the same retail price. However, as each store sets its own retail price based on its assessment of local market conditions, cardigan prices between stores differ. (As all stores are far apart from each other geographically, none competes with any other.)
To understand better what drives the annual number of cardigans sold per store, the CEO estimates the following regression model using 2019 data from 30 of the company’s retail stores:
ln() = 0 + 1ln() + 2 + , (B2.1)
where is the number of cardigans sold by store in 2019, is store ’s 2019 average retail price of cardigans in Australian dollars (AUD), is store ’s expenditure on local advertising in thousands of AUD in 2019, ‘ln’ denotes the natural log, and is an error term. The regression results below were obtained. (Note that ‘LOG’ denotes the natural log.)
Dependent Variable: LOG(Q)
Method: Least Squares
Sample: 1 30
Included observations: 30
Variable Coefficient Std. Error t-Statistic Prob. |
0.55166 0.20824 0.05237 |
(a) Interpret both estimated slope coefficients.
(b) Suppose the CEO wants to test whether 1 < −1. Write out suitable null and alternative
hypotheses for this test and use the regression results above to perform the test at the 5% significance level. State your conclusion.
(c) Considering only the test result obtained in (b) above, do you think that a CardigANZ store could increase its revenue (i.e., × ) by lowering its average retail price by 5% (holding all else constant)? Explain your answer.
(d) Let denote the elasticity of quantity sold with respect to average retail price and let
denote the elasticity of quantity sold with respect to advertising expenditure. Estimate the level of advertising expenditure at which these elasticities sum to zero; i.e.,
+ = 0.
B3. Time series
Consider the time series model,
− = −1 + 1 , = 1, 2, … , , (B3.1)
where is the sample size and 1 is an error term.
(a) If the total multiplier is four times as large as the impact multiplier, what is the
numerical value of the two-period interim multiplier?
(b) If follows the AR(1) process, = + −1 + 2 , where 2 is an error term, and
the fourth autocorrelation coefficient (corr( , −4)) is 0.3, what is the numerical value of (rounded to two decimal places)?
(c) Suppose that () = 2 . Assuming current values of one ( = = 1), use your answers to (a) and (b) above to forecast the dependent variable of (B3.1) one period ahead; i.e., calculate +1 , where the superscript denotes forecast.
(d) Based on the residual correlogram for (B3.1) below, should the forecasting model be improved, and why (or why not)? If it should be improved, how could this be done?
2022-02-15