Page |1

ECO 204/Nov 23rd 2022

[email protected]

ECONOMETRIC (ECO 204) MID-EXAM

STUDENT NAME : CITRA WIDURI
STUDENT ID : 3212210006

1. Suppose I am collecting data survey on:

Sales : sales
Marketing Expense : mark_expense
Price : price
Product Origin : prod_origin

a. Below is the equation to estimate the effects of Marketing Expense, Price and
Product Origin to the sales:

sales = β0 + β1mark_expense + β2price + β3prod_origin + u

where:
sales = annual number of sales (in $)
mark_expense = annual marketing expenditure ($)
price = average prices
prod_origin = binary ( 1 for imported product, 0 for domestic product)
b. Marketing expenses and product origin might give positive returns on sales, but the
average price can be positive or negative depends to the product origin. If the product
is domestic then, the more expensive can be giving less sales, but if the product is
imported, even though it is expensive, it still giving positive sales.

c. Dummy variable is applying to identify product origin, whether it is imported or

domestic ones. We give 1 for the imported because we assume this is contributing to
sales, whilst 0 the domestic product because it is common and didn’t give added value
to the price. But, it still open for vice versa.

2. In simple regression model, unbiased of Ordinary Least Square estimation (i.e. to have
unbiased estimators for the parameters in the population model) requires a set of
assumptions. In addition, it is important to have the least variance and it needs another
assumption.
 Page |2

ECO 204/Nov 23rd 2022

[email protected]

a. The assumptionsin simple regression model (SLR) to have unbiased estimators for
the parameters in the population model are:
Assumption SLR.1 : Linear in Parameters, in the population model, the dependent
variable, Y is related to the independent variable and the error (or disturbance), u as
Y = β0 + β1 X + u , where β0 and β1 are the population intercept and slope,
respectively.
Assumption SLR.2 : Random Sampling, to build the model in SLR.1. we assume that
we have the random sample size n.
Assumption SLR.3 : Sample Variation in the Explanatory Variable. The slope and
intercept estimate are not defined unless we have some sample variation in the
explanatory variables. Say the sample outcomes on X are not all the same value.
Assumption SLR.4 : Zero Conditional Mean. In order to obtain unbiased estimators of
β0 and β1, we need to impose the zero conditional mean assumption. The error u has
an expected value of zero given any value of the explanatory variable. E(u|x) = 0
Assumption SLR.5 : Homoskedasticity. Even though the SLR can be computed under
assumption SLR1-4, but the expressoins would be somewhat complicated. Instead we
add an assumption that is traditional for cross-sectional analysis. This assumption
states that the variance of the unobservable, u , conditional on x, is constant. This is
known as the homoskedasticity or “constant variance” assumption. The error u has
the same variance given any value of the explanatory variable: Var (u|x) = σ2

b. The assumptions should be fulfilled because we need to find the fitted model which
has unbiased estimators for the parameters in the population model, or we can say it
is the minimum variance unbiased estimator, if the assumptions are not fulfilled we
get a model that cannot be used to explain the phenomena.

c. Normality is very important in a simple regression model because to fulfill the

assumption SLR.5, for multiple regression is MLR.5, we include assumption MLR.6,
when we condition on the values of independent variables in our sample, it is clear
that the sampling distributions of the OLS estimators depend on the underlying
distribution of the errors. To make the sampling distribution of the βj tractable, we
assume that the unobserved error is normally distributed in the population.
 Page |3

ECO 204/Nov 23rd 2022

[email protected]

3. We are studying the relationship between salary of employees and company’s sales
revenue and profit, and their experience. We build an econometric model as follows.

𝑠𝑎𝑙𝑎𝑟𝑦 = 𝛽0 + 𝛽1𝑠𝑎𝑙𝑒𝑠 + 𝛽2𝑝𝑟𝑜𝑓𝑖𝑡 + 𝛽3𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒 + 𝑢

Salary and sales are in logarithms, profit in millions of USD, and experience is years.

Below is output of the regression.

The estimation equation is:

𝑠𝑎𝑙𝑎𝑟𝑦 = 5.19668 + 0.2074𝑠𝑎𝑙𝑒𝑠 + 0.0002𝑝𝑟𝑜𝑓𝑖𝑡 – 0.0067𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒+ 𝑢

a. From the equation above, we can interpret that sales and profit are giving positive
returns to the salary but years of experience are giving negative returns, although it
is very low (from 1 year of experience, it only contributes negatively less than 3 days).
If sales revenue increases by 10% it would be giving 10%x0.2074% = 2.074% increase
to the salary. And if the profit increase by USD 100 million, it would be giving
100x0.0002 = 0.02% (USD 1,043.33) increase to the salary. The detail calculation are
below:
 Page |4

ECO 204/Nov 23rd 2022

[email protected]

Sales increases 10% → LSalary = 5.19668% + 2.074% = 7.27068%

Profit increases USD 100millions → Lsalary = 5.19668% + 0.02% = 5.21668%

b. We can reject 𝐻0: 𝛽1 = 0 against 𝐻1: 𝛽1 > 0 at significance level (α) of 5%, when the P
value of 𝛽1 is less than 0.05. And also we can reject the 𝐻0: 𝛽2 = 0 against 𝐻1: 𝛽2 > 0
if the P value of 𝛽2 is less than 0.05.

c. When the null hypothesis, H0: β1 = β2 = β3 = 0, is rejected. the interpretation should

be no linear relationship between the salary of employees and the company’s sales
revenue and profit, and their experience. ( Because if the P value is 0, that means no
significant explanation from the explanatory variables)

d. From the R-squared value 0.3073, we conclude that all of the explanatory variables
can only explain 30.73% of the variance for the dependent variable, and there are
69.27% beyond the equations.

e. I think we should add more variables to the equation and make some testing to the
relationship between each other more, including adding or dropping a few of them,
also we can try to mix the logarithmic for more variables to increase the sensitivity.
We should make some comparisons among those alternative models until we could
find the fitted one.

4. Here is an equation explaining Chief Executive Officers (CEOs) salary (with t statistic in
parentheses below the coefficient):

where finance, transport, and utility are binary variables indicating the financial,
transportation, and utilities industries. The omitted industry is consumer products.
 Page |5

ECO 204/Nov 23rd 2022

[email protected]

a. From the coefficients of binary variables for financial, transportation and utilities
industries, we can read that transportation is the highest contributor factor among
other industries. I mean if you are working in the transportation industry, according
to the equation, It gives you 19.2% returns on the salary, 2.3% higher than the
financial industry and 48.4% higher than utility industries. But if we put gender role
here, we can see that female employees in the financial industry are having the
highest rate of salary (32.4%) , surpassing the transportation industries for 13.2%.

b. Based on two-tailed t score, all of the coefficients are significant in 0.2 significant level,
but the female variable is not significant under 0.1 significant level and below. But if
you are female and working in the financial industry, it is significantly explained how
you could gain 32.4% higher salary than others, even in the 0.5 significant level.
The explanation for each coefficients are below:
• For every 1% increase in the sales, it would give 36.8% increase in the salary
• If the employee is working in the financial industry, it would be having 16.9%
returns to the salary.
• If the employee is working in the transportation industry, it would be having
19.2% returns to the salary.
• If the employee is working in the utility industry, it would have 29.2% less to
the salary.
• In any industry, if you are female, you will get 22.7% less salary, except if you
are in financial industry it would add 32.4% more or 32.4%-22.7% = 9.7% in
total addition to the female employee in financial industry.
c. The percentage difference in estimated salary between females and males who work
in finance industry is :
Log(salary)male = 5.68
Log(salary)female = 5.68 – 0.227 + 0.324 = 5.77
Log(salary)female – Log(salary)male = 5.77 – 5.68 = 0.097

Thus, the female worker in financial industry is having 9.7% higher salary than male
worker

The estimated difference is statistically significant because it is higher than 0.05

significant level.
 Page |6

ECO 204/Nov 23rd 2022

[email protected]

d. A model to test whether Ph.D. graduate Chief Executive Officers (CEOs) have salary
advantages over CEOs with either a master’s or a bachelor’s degree.

Log(CEO𝑠𝑎𝑙𝑎𝑟𝑦) = 𝛽0 + 𝛽1phd_degree + 𝛽3master_degree + 𝑢

Where:
Log(CEO𝑠𝑎𝑙𝑎𝑟𝑦) = salary of CEO in log, to minimize the heteroskedasticity
between unit on salary and binary in the education degree
𝛽0 = constanta
𝛽1 = parameter of those with Ph.D degree
𝛽3 = parameter of those with master degree
phd_degree = binary ( 1 for Ph.D, 0 for others)
master_degree = binary ( 1 for master, 0 for others (bachelor and below)
𝑢 = errors